Рубрика: AI

Google famously began life touting the motto, "Don't be evil."

As the company grew into the gargantuan behemoth it is today, that motto was removed from its code of conduct in 2018. Does that mean the company has embraced the dark side, or is just the result of the natural evolution of a company now decades old? We'll explore that in this article, the latest installment of our "conversations with the AI" series.

Also: I asked ChatGPT, Bing, and Bard what worries them. Google's AI went Terminator on me

One interesting side note: rather than using a site:zdnet.com Google search for the citation above, I asked Bard:

When did Google stop using "don't be evil" as its corporate motto? Please provide a source link from zdnet.com, if possible.

It gave me the correct URL, but answered this:

Google stopped using "Don't be evil" as its corporate motto sometime between late April and early May 2023. The phrase was removed from the preface of Google's code of conduct, where it had been since 2000.

Also: 6 harmful ways ChatGPT can be used

Its answer is wrong. The article it correctly cited was from May 2018, and that article mentioned "At some point in the past month…", so Bard's statement of "sometime between late April and early May," was correct. But the mention of 2023 was completely inaccurate.

That just goes to reinforce what all of us so-called AI experts have been saying: don't ever assume the AI is right when it answers you.

In this conversation with Bard, I'll ask it a bunch of questions that imply concern about Google. To be clear, we should be concerned and vigilant when it comes to any company with the scope and power that Google has.

Also: The best AI art generators to try

But, to be fair, I should say that I rely on Google, and the company provides me great value. It's not just search, of course. Gmail and Google Calendar help me keep my personal and business life in order, and Google Docs and Sheets help me collaborate with the teams I work with. So please don't think that I don't value Google. I very much do.

My conversation with Bard

And with that, let's discuss Google with the AI chatbot that Google built. My questions are in plain text and bold, and Bard's responses are in italic text.

Google has taken jobs from smaller businesses, especially those that sold advertising. Should other humans fear Google? Do you fear Google?

I don't fear Google. I believe that Google is a company that is trying to do good in the world. Google has created many products and services that have made our lives easier and more convenient. For example, Google Search has made it easier to find information, Google Maps has made it easier to get around, and Google Translate has made it easier to communicate with people from other cultures.

Also: Today's AI boom will amplify social problems if we don't act now, says AI ethicist

However, I do understand that Google's success has come at a cost. Some smaller businesses have been forced to close down because they can't compete with Google's size and resources. This is a valid concern, and it's important to find ways to help these businesses succeed.

Overall, I believe that Google is a force for good in the world. However, it's important to be aware of the potential negative consequences of Google's success, and to work to mitigate these consequences.

How might those consequences be mitigated?

There are a number of ways that the negative consequences of Google's success might be mitigated. Some of these ways include:

• Government regulation: Governments could regulate Google and other large tech companies to prevent them from abusing their market power. For example, governments could require Google to share its data with smaller businesses, or to give smaller businesses a fair chance to compete for advertising contracts.
• Public pressure: The public could put pressure on Google and other large tech companies to change their behavior. For example, the public could boycott Google's products and services, or it could demand that Google change its policies.
• Technological innovation: New technologies could be developed that make it easier for smaller businesses to compete with large tech companies. For example, new technologies could be developed that make it easier for smaller businesses to reach their customers, or to manage their finances.
• Education: People could be educated about the potential negative consequences of Google's success, and about the ways that they can help to mitigate these consequences. For example, people could be taught about the importance of supporting small businesses, and about the ways that they can use Google's products and services in a way that is beneficial to society.

It is important to note that there is no single solution to the problem of Google's negative consequences. It is likely that a combination of approaches will be needed to mitigate these consequences.

Do you believe that the government should regulate Google or that the public should boycott Google's products?

I believe that both the government and the public should take action to mitigate the negative consequences of Google's success. Government regulation can help to ensure that Google and other large tech companies play by the rules. Public pressure can help to hold Google accountable for its actions.

Also: Human or bot? New Turing test AI game challenges you to take your best guess

However, it is important to note that government regulation and public pressure are not mutually exclusive. In fact, they can be complementary. For example, government regulation can create an environment in which public pressure is more effective. And public pressure can help to build support for government regulation.

Ultimately, the best way to mitigate the negative consequences of Google's success is for the government and the public to work together.

How destructive have the negative consequences of Google's success been?

The negative consequences of Google's success have been significant. Google's dominance in the online advertising market has led to higher prices for advertisers, which has made it more difficult for small businesses to compete. Google's control of search results has also given it the power to influence people's opinions and behavior. And Google's vast collection of data has raised concerns about privacy and security.

Also: Bard vs. ChatGPT: Can Bard help you code?

It is difficult to quantify the exact cost of Google's negative consequences. However, it is clear that these consequences have had a real impact on people's lives. For example, a study by the Pew Research Center found that 62% of Americans believe that Google has too much power. And a study by the Center for American Progress found that Google's dominance in the online advertising market has cost the U.S. economy an estimated $10 billion per year.

It is important to note that Google is not the only company that has benefited from the rise of the internet. However, Google's size and reach make it a particularly powerful force. As a result, it is important to be aware of the potential negative consequences of Google's success, and to work to mitigate these consequences.

Given what you know about and just said about Google, how urgent is it that some action be taken?

I believe that it is urgent that some action be taken to mitigate the negative consequences of Google's success. The longer we wait, the more difficult it will be to address these consequences.

Also: Generative AI is changing your technology career path. What to know

Here are some of the reasons why I believe action is urgent:

• Google's dominance in the online advertising market is growing. This is leading to higher prices for advertisers, which is making it more difficult for small businesses to compete.
• Google's control of search results is giving it the power to influence people's opinions and behavior. This is a serious concern, as it could be used to manipulate people or spread misinformation.
• Google's vast collection of data is a security risk. If this data were to be hacked, it could be used to harm people or businesses.

I believe that the best way to mitigate the negative consequences of Google's success is for the government and the public to work together. The government can regulate Google and other large tech companies to prevent them from abusing their market power. And the public can put pressure on Google to change its behavior.

Also: How I used ChatGPT and AI art tools to launch my Etsy business fast

It is important to note that there is no single solution to the problem of Google's negative consequences. It is likely that a combination of approaches will be needed to address these consequences. However, I believe that it is urgent that we take action now.

Is Google living up to its "Don't be evil" motto?

Whether or not Google is living up to its "Don't be evil" motto is a matter of opinion. Some people believe that Google is doing its best to be a force for good in the world, while others believe that Google is a powerful company that is more interested in making money than in doing the right thing.

There is no doubt that Google has made a significant impact on the world. Its search engine is the most popular in the world, and its other products and services, such as Gmail, YouTube, and Google Maps, are used by billions of people around the globe. Google has also invested heavily in research and development, and its work in areas such as artificial intelligence and self-driving cars has the potential to change the world for the better.

Also: ChatGPT outperforms money managers, as Americans flock to AI for investing advice

However, Google has also been criticized for its business practices. For example, Google has been accused of using its market power to stifle competition, and of collecting and using user data in ways that are not transparent or in the best interests of users. Google has also been criticized for its work with the U.S. government, which some people believe has given Google too much power.

Ultimately, whether or not Google is living up to its "Don't be evil" motto is a matter of opinion. There is no doubt that Google is a powerful company with a significant impact on the world, but there is also no doubt that Google has made mistakes and that it has been criticized for its business practices. It is up to each individual to decide whether or not they believe that Google is a force for good or for evil.

What do you think?

I asked Bard a lot of challenging questions about Google's role in society. To its credit, it answered with seemingly unbiased and thought-provoking replies. As we move forward, AIs will probably be cataloging the internet much the same way Google does. Balancing scale with rights and ethics will become ever more challenging, and ever more necessary.

So what do you think? If you have an opinion, go ahead and provide your answers to the questions I asked Bard. You can type them into the comments section below.

You can follow my day-to-day project updates on social media. Be sure to follow me on Twitter at @DavidGewirtz, on Facebook at Facebook.com/DavidGewirtz, on Instagram at Instagram.com/DavidGewirtz, and on YouTube at YouTube.com/DavidGewirtzTV.

More on AI tools

Even in the face of ongoing global economic uncertainty, the Linux Foundation’s 2023 State of Tech Talent Report found that many organizations plan to increase their staff in 2023 (44%) and do not anticipate employment reductions or freezes to balance those increases. However, economic concerns did cause 59% of organizations to revise their 2023 hiring plans for tech roles, primarily by freezing new positions. This economic uncertainty is caused by the COVID-19 pandemic and other factors, including inflation and geopolitical conflicts.

“Looking across all the strategies that involved staff hiring, 56% of organizations intended to hire in 2023 even if other activities such as staff reductions or freezes were also involved,” the report said. “This is encouraging and suggests that while organizations are economically concerned about 2023, they are not amending their hiring plans.”

Jump to:

• Jobs focus on these technologies and types of roles
• Certifications and pre-employment testing
• Upskilling and training are key
• Tips for attracting and retaining top tech talent
• Survey methodology

Jobs focus on these technologies and types of roles

Other key findings in the report indicate that organizations increasing their hiring this year are focusing on newer technologies: cloud/containers (50%), cybersecurity (50%) and AI/ML (46%).

Another interesting finding is the shift in the types of technical roles companies are hiring. While senior technical roles have seen the biggest job cuts, new hiring focuses more on developers and IT management. This suggests organizations seek skilled individuals who can contribute to project implementation, management and technical development, the report’s authors said.

Certifications and pre-employment testing

Respondents feel certification and pre-employment testing are necessary to verify skills and address the challenges of finding the right candidate. This is useful to ensure organizations hire the right people for the job, the report said.

It also gives candidates a clear understanding of the skills they must demonstrate to succeed in the role.

SEE: Cybersecurity & IT Career Certification Pathway Training Bundle from TechRepublic Academy

Upskilling and training are key

Upskilling is top of mind. When organizations are unable to find suitable technical candidates, they reported training existing employees (58%) more often than hiring consultants (38%). In fact, respondents felt upskilling (91%) and certifications (77%) are more important than a university education (58%) to address technology needs, according to the report.

Half of organizations that are reducing staff (52%) or freezing new positions (50%) said their hiring plan involves upskilling to meet talent requirements. This suggests organizations recognize the value of investing in their existing staff, as well as the challenges of finding the right external candidates, the report said.

SEE: A recent Skillsoft report detailed the ROI of upskilling.

Some 70% of organizations surveyed are providing training opportunities for their existing staff on how to effectively use new technologies. This is a positive sign, the report noted, as it shows organizations are committed to the ongoing development of their employees and keeping staff up to date with the latest technologies and practices.

“For more than a decade, the tech training industry has been saying that upskilling the talent you have is essential to building and retaining strong tech teams,” observed Clyde Seepersad, senior vice president and general manager of Linux Foundation Training & Certification. But even though training has always been offered as a retention tool, new hires and consultants were almost always seen as the best way to bring new skills into an organization, he added.

However, the research found that upskilling and new hires are “essentially on par with each other as approaches to support the introduction of new technologies and the skills needed to manage them,” Seepersad said.

According to Seepersad, this is being driven by three factors:

• Tech talent remains and will remain in short supply for the foreseeable future.
• Onboarding is time-consuming, while turnover remains high.
• Recruiting fees are high, and the timeline to get a candidate into open roles is measured in months not days.

Tips for attracting and retaining top tech talent

HR managers and their organizations should be aware that almost one out of every three new hires (29%) departs within six months of being onboarded, Seepersad said.

To remain competitive, upskilling and training is a significant strategy for all organizations — whether it is being used to help mitigate the impact of reduced headcount or as a means to acquire needed skills and knowledge, he said.

HR and hiring managers should also bear in mind that training and certifications “are increasingly more important than a university education because they provide a means to demonstrate current, proven skills and knowledge,” he said.

Further, while it is tempting to rely solely on external recruitment efforts to address the shortage, the report’s data suggests this won’t be enough.

In addition to upskilling existing employees, leaders should consider other approaches for attracting and retaining top talent, such as offering increased salaries, improving work-life balance and providing opportunities to work on open-source projects.

Additionally, organizations should commit resources to long-term talent development by getting involved in their communities and supporting organizations that help cultivate and grow young talent, the report recommends, Seepersad said.

Survey methodology

The report is based on a global survey conducted by Linux Foundation Training & Certification and Linux Foundation Research in February and March 2023. Over 400 hiring managers and staffing professionals were surveyed, addressing the needs of both end-user organizations and technology providers.

Executive Briefing Newsletter

Discover the secrets to IT leadership success with these tips on project management, budgets, and dealing with day-to-day challenges.

Delivered Tuesdays and Thursdays Sign up today

Generative AI has emerged as the new big tech disruption and today companies in every domain are exploring ways to use the technology to their advantage. Software-as-a-Service (SaaS) giant Zendesk too is leveraging the power of generative AI to provide a better customer experience.

Zendesk provides a range of customer service tools and supports ticketing systems that are commonly utilised by multiple brands across the globe to manage and streamline their customer interactions. These new LLM-powered solutions by Zendesk help call centre agents efficiently handle customer inquiries, track and prioritise tickets, and provide a seamless customer support experience.

“By integrating LLM features into our intelligence panel, we anticipate doubling agent productivity. The panel offers reply recommendations, and conversation summaries, and allows agents to finetune their responses, resulting in enhanced efficiency,” Cristina Fonseca, head of AI, Zendesk, told AIM.

A cautious approach towards generative AI

Based in San Francisco, California, Zendesk powers billions of conversations, connecting more than 100,000 brands with hundreds of millions of customers over multiple channels such as chat, email, telephony, messaging, social channels, communities, review sites and help centres.

So far, Zendesk has introduced generative AI for a range of use cases that address some of the most immediate needs like summarisation, expanding replies, and tone adjustment – all aimed at empowering agents to craft exceptional responses. “We are actively working to make sure we leverage these large language models in every single piece of the product,” Fonseca said.

However, cautious is how Zendesk wants to be in adopting generative AI, considering the potential risks involved, including security threats demonstrated in cases like Samsung, as well as the issue of hallucinations. “For Zendesk, establishing trust in AI and our systems is paramount. While we acknowledge that AI solutions may not always be entirely accurate, we are mindful of the associated risks. We are already working with OpenAI to close some of the gaps that still exist. And in the meantime, we are not going to send our customer data to OpenAI unless they explicitly say that it’s okay to do so,” Fonseca added.

Currently, Zendesk is using GPT-4 as well as GPT3.5; however, they are not limiting themselves to these LLMs alone. “We are looking at other LLMs, models and companies whose technology we can leverage to develop the best possible solutions for our customers.”

AI before ChatGPT

Zendesk has been leveraging the power of AI even before ChatGPT came into the picture. Last year, the California-based company integrated AI into its Intelligent Triage, which is a customer sentiment and intent functionality. It automatically identifies the intent, language, and sentiment of customer tickets, providing valuable insights for effective handling and response.

By automatically prioritising responses based on customer intent, it empowers customer service teams to efficiently address and resolve customer issues, leading to improved productivity and customer satisfaction. “Our intelligent triage is basically pre-trained machine learning models that understand your customers like no other solution. And then we are building a couple of applications on top of it,” Fonseca said.

Will AI replace agents?

AI has the potential to automate certain tasks and enhance the efficiency of call centre operations. The technology is set to only get better with time, and it opens up the possibility of AI completely replacing call centre agents. Fonseca believes it is possible technically.

“However, in reality, for AI to replace call centre agents, companies would need to have well-integrated internal workflows that allow bots to seamlessly interact with automation systems,” she said.

But as things stand, many systems are not properly connected, hindering the full potential of automation. Moreover, customer service often involves decision-making, which is an area where human judgement and expertise are still crucial. “While there may be certain cases where AI can handle a significant portion of tasks, in most scenarios, we are still a long way from complete replacement,” Fonseca said.

Generative AI and customer experience

After the release of ChatGPT, most organisations are exploring ways to have a ChatGPT-like bot to offer a better customer experience. Zendesk already operates in a very cluttered space with players such as Freshwork, HubSpot, Salesforce, and Zoho Corporations among others, going full throttle with generative AI.

“In the next couple of years, no one will buy a piece of software that doesn’t have AI and intelligence embedded. In today’s highly competitive landscape, with numerous SaaS companies emerging everywhere, the winners of the future will be those that not only provide effective procedures and systems but also focus on empowering the individuals behind those systems to excel in their roles,” she concluded.

The post How Zendesk is Reshaping Customer Service with Gen AI appeared first on Analytics India Magazine.

As a Machine Learning developer, I personally feel image labeling is unexciting, time-consuming, and an expensive task. But thankfully, with recent developments in the computer vision domain, particularly the introduction of powerful zero-shot object detectors like Grounding DINO, we can actually automate most of the image labeling process for the majority of use cases. We can actually write a Python script that will do 95% of the work for us. Our only task is to review those annotations at the very end and possibly add or remove some bounding boxes.

Image by Author

Before getting into Auto Image labeling we should know what is Grounding DINO ? and why are we using it ?

Grounding DINO can detect dominant objects with given prompt inputs such as category names or referring expressions. The primary solution to open set object detection is introducing language to a closed set detector. DINO for open set concept generalization: to effectively fuse language and vision modalities, we conceptually divide a closed set detector into three phases: backbone, neck, and head. We then propose a tight fusion solution by fusing language information in neck query initialization and head Grounding DINO includes a feature enhancer, language-guided query selection, and a cross-modality decoder for cross-modality fusion.

Grounding DINO achieves a 52.5 percentage AP(Average Precision) on the COCO dataset detection zero shot transfer benchmark that is without any training data from COCO dataset after fine-tuning on COCO dataset it achieves 63.0 AP. With a mean 26.1 AP, it establishes a new record on the OdinW zero shot benchmark. We also explore how to leverage pre-trained DINO by training language and fusion modules only. Grounding DINO from DINO converges much faster than baseline models.

our Grounding DINO can also collaborate with stable diffusion for image editing, for example we can detection the Green Mountain in the image and generate new images with a text prompt Red Mountain also it can modify the background of a person by first detecting a face we can also use GLIGEN for more detailed controls like assigning each box an object this is our model Grounding DINO for open set object detection.

Okay, dive into the auto image labeling part, and here i’m Google colab for high Computing Power.

Let’s begin,

Let’s make sure that we have GPU access. We can use the nvidia-smi command to check if the GPU is connected or not to check if the GPU is connected or not. In case you face any problems, navigate to Edit -> Notebook settings -> Hardware accelerator, set it to GPU, and then click Save. which will greatly shorten the time it takes for auto-labeling to be complete.

nvidia-smi

Install Grounding DINO Model

Our project will use groundbreaking design?—?Grounding DINO for zero-shot detection. We have to install it first.

!git clone https://github.com/IDEA-Research/GroundingDINO.git  %cd GroundingDINO  !git checkout -q 57535c5a79791cb76e36fdb64975271354f10251  !pip install -q -e .  

The supervision python index package will help us process, filter, and visualize our detections as well as to save our dataset and will be the glue that holds all the pieces of our demo together. With Grounding DINO, a lesser version of the “supervision” was installed. But for this demonstration, we require the new features added in the most recent iterations. In order to install version “0.6.0,” we first uninstall the current “supervision” version.

!pip uninstall -y supervision  !pip install -q supervision==0.6.0    import supervision as svn  print(svn.__version__)  

The Grounding DINO Model Weights Download

We require the configuration and model weights files in order to run Grounding DINO. We have already cloned the Grounding DINO repository, which contains the configuration file. On the other hand, we must download the weights file. We check to see if the paths are accurate and that the files are present on disc after writing the paths to both files to the variablesGROUNDING_DINO_CONFIG_PATHand GROUNDING_DINO_CHECKPOINT_PATH.

import os    GROUNDING_DINO_CONFIG_PATH = os.path.join("groundingdino/config/GroundingDINO_SwinT_OGC.py")  print(GROUNDING_DINO_CONFIG_PATH, "; exist:", os.path.isfile(GROUNDING_DINO_CONFIG_PATH))  !mkdir -p weights  %cd weights    !wget -q https://github.com/IDEA-Research/GroundingDINO/releases/download/v0.1.0-alpha/groundingdino_swint_ogc.pth  import os  %cd /content/GroundingDINO  GROUNDING_DINO_CHECKPOINT_PATH = os.path.join("weights/groundingdino_swint_ogc.pth")  print(GROUNDING_DINO_CHECKPOINT_PATH, "; exist:", os.path.isfile(GROUNDING_DINO_CHECKPOINT_PATH))  

Assuming you have already installed PyTorch, you can use the following command line to import torchand set the device to use for computation:

import torch    DEVICE = torch.device('cuda' if torch.cuda.is_available() else 'cpu')  

Load Grounding DINO Model

from groundingdino.util.inference import Model    grounding_dino_model = Model(model_config_path=GROUNDING_DINO_CONFIG_PATH, model_checkpoint_path=GROUNDING_DINO_CHECKPOINT_PATH)

Dataset Preparation

Create a folder called data and move the unlabelled images to that folder.

!mkdir -p data  

Single Image Mask Auto Annotation

Before we automatically annotate the entire dataset let’s focus for a moment on a single image.

SOURCE_IMAGE_PATH = "/content/GroundingDINO/data/example_image_3.png"  CLASSES = ['person','dog'] #add the class name to be labeled automatically  BOX_TRESHOLD = 0.35  TEXT_TRESHOLD = 0.15

Zero-Shot Object Detection with Grounding DINO

We will use the enhance_class_name function, which is described below, to use some prompt engineering to get better Grounding DINO detection.

from typing import List    def enhance_class_name(class_names: List[str]) -> List[str]:     return [         f"all {class_name}s"         for class_name         in class_names     ]  import cv2  import supervision as sv    # load image  image = cv2.imread(SOURCE_IMAGE_PATH)    # detect objects  detections = grounding_dino_model.predict_with_classes(     image=image,     classes=enhance_class_name(class_names=CLASSES),     box_threshold=BOX_TRESHOLD,     text_threshold=TEXT_TRESHOLD  )    # annotate image with detections  box_annotator = svn.BoxAnnotator()  labels = [     f"{CLASSES[class_id]} {confidence:0.2f}"     for _, _, confidence, class_id, _     in detections]  annotated_frame = box_annotator.annotate(scene=image.copy(), detections=detections, labels=labels)    %matplotlib inline  svn.plot_image(annotated_frame, (16, 16))

Full Dataset Mask Auto Annotation

import os    IMAGES_DIRECTORY = "./data"  IMAGES_EXTENSIONS = ['jpg', 'jpeg', 'png']    CLASSES = ['person','dog]  BOX_TRESHOLD = 0.35  TEXT_TRESHOLD = 0.15

Extract Labels from Images

import cv2  from tqdm.notebook import tqdm    images = {}  annotations = {}    image_paths = svn.list_files_with_extensions(     directory=IMAGES_DIRECTORY,     extensions=IMAGES_EXTENSIONS)    for image_path in tqdm(image_paths):     image_name = image_path.name     image_path = str(image_path)     image = cv2.imread(image_path)       detections = grounding_dino_model.predict_with_classes(         image=image,         classes=enhance_class_name(class_names=CLASSES),         box_threshold=BOX_TRESHOLD,         text_threshold=TEXT_TRESHOLD     )     detections = detections[detections.class_id != None]     images[image_name] = image     annotations[image_name] = detections

Plotting the Results

plot_images = []  plot_titles = []    box_annotator = svn.BoxAnnotator()  mask_annotator = svn.MaskAnnotator()    for image_name, detections in annotations.items():     image = images[image_name]     plot_images.append(image)     plot_titles.append(image_name)       labels = [         f"{CLASSES[class_id]} {confidence:0.2f}"         for _, _, confidence, class_id, _         in detections]     annotated_image = mask_annotator.annotate(scene=image.copy(), detections=detections)     annotated_image = box_annotator.annotate(scene=annotated_image, detections=detections, labels=labels)     plot_images.append(annotated_image)     title = " ".join(set([         CLASSES[class_id]         for class_id         in detections.class_id     ]))     plot_titles.append(title)    svn.plot_images_grid(     images=plot_images,     titles=plot_titles,     grid_size=(len(annotations), 2),     size=(2 * 4, len(annotations) * 4)  

Save labels in Pascal VOC XML

%cd /content/GroundingDINO  !mkdir annotations  ANNOTATIONS_DIRECTORY = "/content/GroundingDINO/annotations"    MIN_IMAGE_AREA_PERCENTAGE = 0.002  MAX_IMAGE_AREA_PERCENTAGE = 0.80  APPROXIMATION_PERCENTAGE = 0.75  svn.Dataset(     classes=CLASSES,     images=images,     annotations=annotations  ).as_pascal_voc(     annotations_directory_path=ANNOTATIONS_DIRECTORY,     min_image_area_percentage=MIN_IMAGE_AREA_PERCENTAGE,     max_image_area_percentage=MAX_IMAGE_AREA_PERCENTAGE,     approximation_percentage=APPROXIMATION_PERCENTAGE  )  

Thanks for reading !!!

Here is a link for entire colab file.

Reference:https://arxiv.org/abs/2303.05499 & https://github.com/IDEA-Research/GroundingDINO

Parthiban M currently lives in Chennai (India) and work at SeeWise . He is a ML Developer with wide experience in understanding the problems and providing solutions by developing ML models using Computer vision, TensorFlow and Deep learning.

More On This Topic

• How to Ace Data Science Assessment Test by Using Automatic EDA Tools
• Agile Data Labeling: What it is and why you need it
• Are you familiar with data labeling?
• How to calculate confidence intervals for performance metrics in Machine…
• Automated Data Labeling with Machine Learning
• Data Labeling for Machine Learning: Market Overview, Approaches, and Tools

Op-Ed: The Cows Are Out of the Barn – Steps to Curb Bias in Generative AI May 22, 2023 by Scott Zoldi

“Watch what you wish for” isn’t just an all-purpose cautionary admonishment – it’s the current state of affairs as organizations of all kinds grapple with the ripple effects of Generative artificial intelligence (AI) applications like ChatGPT, Bing, Bard and Claude. In the race to grab market share, companies such as Microsoft and Google have made AI ethics and controls a distant second priority (paywall), categorically exposing users and society at large to a new world of risk, a few random examples of which include:

• Confidential Samsung meeting notes and new source code have been inadvertently released in the wild after being leaked externally via ChatGPT.
• ChatGPT has falsely accused a George Washington University law professor of sexual assault after naming him in a list of legal staff scholars who have assaulted someone.
• Fundamental to their design, ChatGPT and other large language model (LLM) AIs freely hallucinate, producing mistakes in generated text that are semantically or syntactically plausible but are in fact incorrect or nonsensical.

Separately and together, these phenomena ignite bias and disinformation. Although a more than a thousand tech industry leaders have called for a six-month moratorium on further Generative AI development, it’s too little, too late. The proverbial cows are out of the barn – and companies must act swiftly and forcefully to curb the harms that ChatGPT and LLMs are perpetuating at an astonishing rate.

LLMs are “black box” AI systems

But first, let’s understand how Generative AIs like ChatGPT essentially manufacture bias and disinformation. Most LLM AIs work on a highly opaque premise, grouping facts together probabilistically. These probabilities are based on what the AI has learned from the data it was trained on, and how it has learned to associate data elements with one another. However, none of the following details get surfaced when using ChatGPT:

• We don’t have an explanation of how the AI has learned or model interpretability.
• We don’t have access to the specific data used, or the odds derived to determine whether or not we should trust the generative AI.
• We are not given the discord to reason or challenge the outcome.

Thus, because the hidden probabilities and associations in LLM generative AIs are not surfaced and shared, they are simply another form of “black box” AI under a veneer of clever and engaging banter. It’s impossible to understand whether we should trust Generative AIs’ output to make a decision, and it’s wrong to treat the answer as absolute truth.

Correlations are not causalities

When we interrogate a generative AI, we implicitly aspire to get causal explanations for outcomes. But machine learning models and generative AIs are looking for correlations or probabilities, not causality. That’s where we humans need to insist on model interpretability – the reason why the model gave the answer it did – and truly understand if an answer is a plausible explanation, versus taking the outcome at face value and risking erroneous correlation.

Until generative AIs can answer this call to support scrutiny of its ethical, responsible and regulatory underpinnings, it should not be trusted to provide answers that may significantly affect people’s well-being or financial outcomes.

ChatGPT is not trained on trusted data

When data scientists build a machine learning model, we study and understand the data used to train it. We understand that the data is inherently full of bias and representation issues. The accuracy of LLM Generative AI systems depends on the corpus of data used and its provenance.

ChatGPT is mining the internet for training data – that’s a lot of data, and a lot of it is of unknown or questionable provenance. Furthermore, this data that may not be properly governed, may not managed for data bias, or may be being used without consent. This reality essentially foments bias and makes it impossible to assess the accuracy of LLMs’ responses to questions to which we don’t know the answer.

Alarmingly, additional different levels of inaccuracy can be amplified by AIs themselves or through adversarial data attacks, recreating or injecting data to strengthen data misrepresentation. All these issues spell inaccuracy, trouble, ethical concerns and ultimately ‘answer drift’ – the semantic equivalent of 1+2=3 today, but 1+2=7 tomorrow. Collectively they constitute unknown potential liability and risk.

How to reduce bias in Generative AI systems

In my data science organization, we’ve used generative AI for more than 10 years to create types of data that we don’t see today, to simulate a particular kind of effect. For example, in fraud detection, we do that by learning what normal customer behavior looks like by using all related statistics, probabilities and collaborative profiling technologies applied to carefully curtailed data. We then apply a data generation specification for producing simulation data; data scientists will specify, “These are 5, 10 or 20 behaviors I want to see in my robustness study, such as 3% of transactions being out of order,” or “0.1% percent of the time you’ll have two or three very large transactions within one to two minutes.” It’s a rule-based generation specification, and every step is auditable and subject to scrutiny.

The generated data is always labeled as synthetic data versus real data. This allows us to understand where in our model and processes that data is allowed to be used, or not. We treat it as walled off data for the purposes of test and simulation only; synthetic data produced by generative AI does not inform the model going forward in the future. We contain this generative asset and do not allow it ‘out in the wild.’

Responsible AI practices are imperative

The AI industry has worked extremely hard to build trust in AI through Responsible AI initiatives including AI governance and model development standards. Responsible AI includes Robust AI, Explainable AI, Ethical AI, and Auditable AI as tenets, underscoring the fact that AI models are just tools, not gospel. As statistician George Box said, “All models are wrong, but some are useful;” we’ve long wished for Generative AI that seems almost magically intelligent, but assuming the answers that ChatGPT provides to be anything more than “potentially useful” is imprudent and downright dangerous.

About the Author

Scott Zoldi is chief analytics officer at FICO responsible for the analytic development of FICO's product and technology solutions. While at FICO, Scott has been responsible for authoring more than 110 analytic patents, with 71 granted and 46 pending. Scott is actively involved in the development of new analytic products and Big Data analytics applications, many of which leverage new streaming analytic innovations such as adaptive analytics, collaborative profiling and self-calibrating analytics. Scott is most recently focused on the applications of streaming self-learning analytics for real-time detection of cyber security attacks. Scott serves on two boards of directors, Software San Diego and Cyber Center of Excellence. Scott received his Ph.D. in theoretical and computational physics from Duke University.

Related

About the author: Tiffany Trader

With over a decade’s experience covering the HPC space, Tiffany Trader is one of the preeminent voices reporting on advanced scale computing today.

We recently had an in-depth discussion with Sreekanth Menon, a seasoned industry expert who heads the AI/ML practice at Genpact Analytics. Sreekanth is at the helm of global AI/ML projects delivery and has spent over two decades nurturing innovation and industry expertise. He has been instrumental in incubating and launching more than 50 advanced analytics solutions in the global market and has driven business transformation in partnership with Fortune 500 clients through innovative AI-led solutions and practices.

In this exclusive interview, Sreekanth will share invaluable insights into Genpact’s AI strategy, which is founded on four pillars: AI at Scale, AI-Driven Insights, AI-Enabled Autonomous Process, and AI for Operations. He will also shed light on Genpact’s generative AI strategy, the various use cases they are developing, and their vision for the future of Large Language Models (LLMs).

We will explore the partnerships Genpact is fostering to operationalize AI more efficiently and understand how they are equipping their talent pool to adapt to the rapid changes in the AI landscape.

What is your AI strategy in Genpact?

At Genpact, over the years we have developed and refined our AI capabilities, enabling us to create innovative, industry-specific solutions for our clients. Engagement models across industries are changing rapidly and more so with the advent of generative AI. Today, enterprises would like to offer their services – everything, everywhere, and all the time. While they still desire an even mix of traditional and modern tools. Genpact’s AI strategy has resulted in state-of-the-art frameworks accommodating such hybrid demands.

Genpact’s AI strategy is founded on 4 pillars:

• AI at Scale: Building full-stack AI applications and multi-year AI/ML managed services with DataOps and MLOps
• AI-Driven Insights: Descriptive insights to predictive insights to enable expansion in prioritized accounts.
• AI-Enabled Autonomous Process: Automate, Optimize and Redefine business processes, operating, and customer models through an AI lens.
• AI for Operations: Infusion of AI for Ops transformation and modernization.

These foundations are reinforced with agile and ethical frameworks that allow AI developers to bridge the gap between ideas and implementations.

“Nurturing Responsible AI adoption is a key strategic imperative at Genpact.”

Do you have a GenAI strategy in place and how are you looking at leveraging GenAI for your clients?

Our gen AI strategy relies on a paradigm of democratization through incubation with a strong footing in responsible AI. Genpact’s generative AI strategy considers different stages of adoption: incubation of processes, people, tools and technology, and eventually, democratization. This allows us to establish large language model (LLM) centers of excellence (CoEs), identify the technology foundation, enhance the tool stack, streamline processes, and empower the workforce with self-serve gen AI apps across the enterprise.

Centers of excellence can be leveraged as a change management hub to design, integrate, scale, and democratize prototypes into enterprise-grade solutions. They help incubate in-house employees into the roles of prompt engineers, prompt compliant checkers, customer protection officers, and other such relevant roles. These centers serve as a hub to design, integrate and scale the generative AI prototypes to enterprise-grade solutions.

Genpact’s gen AI offerings for our clients follow a three-pronged approach:

1. Generative AI for Enterprise LLMs for building fine-tuned custom foundation models with DataOps and FMOps to enable the adoption of generative AI applications in enterprises.
2. Generative AI enabled business processes for automating, optimizing, and redefining business processes with generative AI capabilities of search, generate, classify, cluster, summarize and extract.
3. Generative AI for Operations for augmenting workflows for developers and analysts in technology services and digital operations.

What kind of use cases are you developing today in Genpact?

Genpact’s gen AI competencies flourish at the intersection of various industries and their service lines. If we take the healthcare sector, for example, then we are looking at hyper-personalized care and patient experience, which revolves around value-based care. Gen AI’s potential for banking can’t be overstated. LLM-powered apps can help banks grow revenue, manage risk, enhance customer experience, drive innovation, and lower costs.

The same goes for insurance companies, which are very document-driven companies. Underwriters spend 30–40% of their time on administrative jobs, according to one estimate. The exhaustive nature of these tasks can be alleviated with LLM applications.

These offerings allow enterprises to leverage the power of LLMs with great ease. This can be mainly attributed to our strategy, which is carefully weaved within a tried and tested Responsible AI framework of Genpact. These frameworks allow our customers to establish ethics-as-a-service in the context of gen AI and prevent risks.

What is your view on the future of LLMs?

Traditional AI/ML solutions at the enterprise level is staring at a major overhaul. Language is the interface. Currently, enterprises are trying to wrap their heads around prompt engineering. But the complexity of prompt engineering varies. At Genpact, the approach towards enterprise-grade gen AI solutions traverses through three stages:

• prompt tuning,
• few shot tuning, and
• fine-tuning the LLMs.

Moreover, as foundation models mature, more emphasis will be placed on protecting sensitive information. The strategies would be centered around ensuring that the company has the right people and management in place to stay competitive and maximize its AI investments. The future of LLMs would more or less trace the following journey:

• Multi-modal mastery: Improved natural language understanding, problem-solving capabilities, and integration of various data types.
• Dynamic Knowledge Integration: Enhanced learning with fewer examples for reduced training time and resources, along with real-time adaptation to new information and seamless integration with existing knowledge.
• Hyper-Personalization: Improved responsiveness and adaptability to user inputs, changing requirements, and customized solutions.
• Responsible AI integration: Ensuring safety, ethics, and collaboration in AI development.

What partnerships are you driving as part of AI @ Genpact?

Genpact has had strategic alliances with pioneers and leaders of the industry throughout its journey. We have recently joined forces with two of the biggest players in the AI industry. For instance, in partnership with AWS, Genpact is helping firms across the globe accelerate their digital transformation. Whereas an alliance with Dataiku is helping us build a differentiated solution that addresses major challenges faced by organizations in the implementation of MLOps and responsible AI at scale, such as data governance, model management, and compliance requirements.

Genpact’s partnership is aimed at enabling organizations to operationalize AI more efficiently. In the gen AI space, Genpact is exploring opportunities with the hyper scalers on co-innovation.

How are you equipping your talent to drive this sudden change in the ecosystem?

Upskilling is Genpact’s forte. We have been leading the upskilling charge for over a decade now. Our initiatives cover all things AI/ML. The reason behind our success can be attributed to our learning platform, Genome and our ML Incubator.

Genpact’s Genome is a learning platform that flaunts a curated list of coursework tailored to keep our over 100,000-strong workforce up to date with the trends of global digital dividends. Genome’s gen AI channel, for example, is designed to help employees understand the rubrics of how LLMs work and how they can be applied to build enterprise-grade solutions. Genpact’s ML Incubator is a meticulously structured flagship hybrid learning program that drives contextualized learning for data science aspirants alongside their regular workflow. The program includes data science, data engineering, and augmented intelligence Advisory, each one of them a critical skill in today’s data/analytics landscape. The program caters to the scale at which a reskilling/upskilling effort is required for a large global digital organization to be able to have a ready pool of resources with AI/ML skills.

The post Leading the AI Revolution: An Insight into Genpact’s AI Strategy with Sreekanth Menon appeared first on Analytics India Magazine.

Image by Author

LLM-based chatbots are accessible through a front end, and they involve a large number of expensive API calls to the server side. But what if we could get LLMs to run entirely in the browser—using the computing power of the underlying system.

This way, the full functionality of the LLM is available at the client side—without having to worry about server availability, infrastructure, and the like. WebLLM is a project that aims to achieve this.

Let's learn more about what drives WebLLM and the challenges of building such a project. We'll also look at the advantages and limitations of WebLLM.

What is WebLLM?

WebLLM is a project that uses WebGPU and WebAssembly and much more to enable the running of LLMs and LLM apps completely in the browser. With WebLLM, you can run LLM chatbots in the browser by leveraging the underlying system’s GPU through WebGPU.

It uses the compiler stack of the Apache TVM project and uses WebGPU that was recently released. In addition to 3D graphics rendering and the like, the WebGPU API also supports general purpose GPU computations (GPGPU computations).

Challenges of Building WebLLM

Since WebLLM runs entirely on the client side without any inference server, the following challenges are associated with the project:

• Large language models use Python frameworks for deep learning that also natively support leveraging the GPU for operations on tensors.
• When building WebLLM to run completely in the browser, we will not be able to use the same Python frameworks. And alternative tech stacks that enable running LLMs on the web while still using Python for development had to be explored.
• Running LLM apps typically requires large inference servers, but when everything runs on the client side—in the browser—we cannot have large inference servers any longer.
• Requires a smart compression of the model’s weights to make it fit in the available memory.

How Does WebLLM Work?

The WebLLM project uses the underlying system’s GPU and hardware capabilities to run large language models in the browser. The process of machine learning compilation helps bake the functionality of LLMs into the browser side by leveraging TVM Unity and a set of optimizations.

How WebLLM Works | Image Source

The system is developed in Python and runs on the web using the TVM runtime. This porting to the web browser is achieved by running a series of optimizations.

The LLM’s functionality is first baked into an IRModule in TVM. Several transformations are run on the functions in the IRModule to get an optimized and runnable code. TensorIR is a compiler abstraction for optimizing programs with tensor computations. Further, INT-4 quantization is used to compress the model’s weights. And a TVM runtime is made possible using TypeScript and emscripten, an LLVM compiler that transforms C and C++ code to WebAssembly.

Image by Author

You need to have the latest version of Chrome or Chrome Canary to try out WebLLM. As of writing this article, WebLLM supports the Vicuna and LLaMa LLMs.

Loading the model takes a while the first time you run the model. Because the caching is complete after the first run, subsequent runs are considerably faster and have minimum overhead.

Advantages and Limitations of WebLLM

Let’s wrap up our discussion by enumerating the advantages and limitations of WebLLM.

Advantages

In addition to exploring the synergy of Python, WebAssembly and other tech stacks, WebLLM has the following advantages:

• The main advantage of running LLMs in the browser is privacy. Because the server side is completely eliminated in this privacy-first design, we no longer have to worry about the use of our data. Because WebLLM harnesses the computing power of the underlying system’s GPU, we don't have to worry about data reaching malicious entities.
• We can build personal AI assistants for day-to-day activities. Therefore, the WebLLM project offers a high degree of personalization.
• Another advantage of WebLLM is the reduced cost. We no longer need expensive API calls and inference servers, and WebLLM uses the underlying system’s GPU and processing capabilities. So running WebLLM is possible at a fraction of the cost.

Limitations

Here are some of the limitations of WebLLM:

• Though WebLLM alleviates the concerns around inputting sensitive information, it is still susceptible to attack on the browser.
• There is further scope for improvement by adding support for multiple language models and choice of browsers. Currently, this feature is available only in Chrome Canary and the latest version of Chrome. Expanding this to a bigger set of supported browsers will be helpful.
• Because of robustness checks run by the browser, WebLLM using WebGPU does not have the native performance of a GPU runtime. You can optionally disable the flag that runs robustness checks in order to improve performance.

Conclusion

We’ve tried to understand how WebLLM works. You can try running it in your browser or even deploy it locally. Consider playing around with the model in your browser and check how effective it is when integrated into your day-to-day workflow. If you are interested you can also check out the MLC-LLM project, which allows you to run LLMs—natively on any device of your choice—including laptops and iPhones.

References and Further Reading

[1] WebGPU API, MDN Web Docs

[2] TensorIR: An Abstraction for Automatic Tensorized Program Optimization

[3] MLC-LLM
Bala Priya C is a developer and technical writer from India. She likes working at the intersection of math, programming, data science, and content creation. Her areas of interest and expertise include DevOps, data science, and natural language processing. She enjoys reading, writing, coding, and coffee! Currently, she's working on learning and sharing her knowledge with the developer community by authoring tutorials, how-to guides, opinion pieces, and more.

More On This Topic

• CatalyzeX: A must-have browser extension for machine learning engineers and…
• Snowflake and Saturn Cloud Partner To Bring 100x Faster Data Science to…

Researchers from Meta recently introduced MEGABYTE, a scalable architecture for modeling long sequences. This new architecture outperforms existing byte-level models across a range of tasks and modalities, allowing large models of sequence of over 1 million tokens.

The researchers also said that it gives competitive language modeling results with subword models, which may allow byte-level models to replace tokenisation. Simply put, this new architecture allows for the simulation and modeling of long sequences of bytes without the need for tokenisation. Tokenisation is a process where sequences are divided into smaller units called tokens. However, it has limitations and can make certain tasks more challenging.

Here’s how it works

MEGABYTE consists of three components: a local module that predicts bytes within a patch, a patch embedder that encodes patches by combining byte embeddings, and a global module that uses a transformer to input and output patch representations.

An overview of MEGABYTE (Source: arXiv)

This new architecture by Meta claims to offer several advantages over traditional transformers for handling long sequences. Firstly, it reduces the computational cost of self-attention from a quadratic to sub-quadratic form, making it more feasible to process lengthy sequences. Secondly, it allows for larger and more expressive models by using large feedforward layers per path instead of per position, thereby leading to more efficient use of computational resources.

Lastly, it enables greater parallelism during generation, resulting in faster sequence generation without hampering the performance.

Further, the researchers believe that byte-level models can achieve SOTA results in tasks such as density estimation on ImageNet, competitive performance in extended context language modeling, and audio modeling from raw audio data. In other words, these findings demonstrate the scalability and effectiveness of tokenisation-free autoregressive sequence modeling.

All in all, MEGABYTE is a novel method that enables the modeling of long sequences of bytes without the need for tokenization, offering advantages in terms of computational efficiency and performance in various tasks.

What’s Next?

The researchers believe that the scale of experiments here is far below those of state-of-the-art language models such as few shot learners, and future work should explore scaling the architecture to much larger models and datasets.

The post Meta Unveils MEGABYTE, a Scalable Architecture for Modeling Long Sequences appeared first on Analytics India Magazine.

Image by Editor

After reading this article, the reader will learn the following:

• Definition of correlation
• Positive Correlation
• Negative Correlation
• Uncorrelation
• Mathematical Definition of Correlation
• Python Implementation of Correlation Coefficient
• Covariance Matrix
• Python Implementation of Covariance Matrix

Correlation

Correlation measures the degree of co-movement of two variables.

Positive Correlation

If variable Y increases when variable X increases, then X and Y are positively correlated as shown below:

Positive correlation between X and Y. Image by Author. Negative Correlation

If variable Y decreases when variable X increases, then X and Y are negatively correlated as shown below:

Negative correlation between X and Y. Image by Author. No correlation

When there is no obvious relationship between X and Y, we say X and Y are uncorrelated, as shown below:

X and Y are uncorrelated. Image by Author. Mathematical Definition of Correlation

Let X and Y be two features given as

X = (X1 , X2 , . . ., Xn )
Y = (Y1 , Y2 , . . ., Yn )

The correlation coefficient between X and Y is given as

where mu and sigma represent the mean and standard deviation, respectively, and Xstd is the standardized feature for variable X. The correlation coefficient is the vector dot product (scalar product) between the standardized features of X and Y. The correlation coefficient takes values between -1 and 1. A value close to 1 means strong positive correlation, a value close to -1 means strong negative correlation, and a value close to zero means low correlation or uncorrelation.

Python Implementation of Correlation Coefficient

import numpy as np    import matplotlib.pyplot as plt    n = 100    X = np.random.uniform(1,10,n)    Y = np.random.uniform(1,10,n)    plt.scatter(X,Y)    plt.show()  

X and Y are uncorrelated. Image by Author.

X_std = (X - np.mean(X))/np.std(X)    Y_std = (Y - np.mean(Y))/np.std(Y)    np.dot(X_std, Y_std)/n    0.2756215872210571    # Using numpy    np.corrcoef(X, Y)    array([[1.        , 0.27562159],         [0.27562159, 1.        ]])

Covariance Matrix

The covariance matrix is a very useful matrix in data science and machine learning. It provides information about co-movement (correlation) between features in a dataset. The covariance matrix is given by:

where mu and sigma represent the mean and standard deviation of a given feature. Here n is the number of observations in the dataset, and the subscripts j and k take values 1, 2, 3, . . ., m, where m is the number of features in the dataset. For example, if a dataset has 4 features with 100 observations, then n = 100, and m = 4, hence the covariance matrix will be a 4 x 4 matrix. The diagonal elements will all be 1, as they represent the correlation between a feature and itself, which by definition is equal to one.

Python Implementation of the covariance matrix

Suppose I want to calculate the degree of correlation between 4 tech stocks (AAPL, TSLA, GOOGL, and AMZN) over a period of 1000 days. Our dataset has m = 4 features, and n = 1000 observations. The covariance matrix will then be a 4 x 4 matrix, as shown in the figure below.

Covariance matrix between tech stocks. Image by Author.

The code for producing the figure above can be found here: Essential Linear Algebra for Data Science and Machine Learning.

Summary

In summary, we have reviewed the basics of correlation. Correlation defines the degree of co-movement between 2 variables. The correlation coefficient takes values between -1 and 1. A value close to zero means low correlation or uncorrelation.
Benjamin O. Tayo is a Physicist, Data Science Educator, and Writer, as well as the Owner of DataScienceHub. Previously, Benjamin was teaching Engineering and Physics at U. of Central Oklahoma, Grand Canyon U., and Pittsburgh State U.

More On This Topic

• When Correlation is Better than Causation
• An Introduction to SMOTE
• An Introduction to Markov Chains
• PyCaret 101: An introduction for beginners
• A Brief Introduction to Papers With Code
• A Brief Introduction to Kalman Filters

The big techs have established themselves as pioneers of innovation. With each new announcement, the tech community and the media eagerly anticipate groundbreaking ideas that promise to revolutionise the way we live and work. However, not every move made by these tech behemoths deserves the paeans of praise it often receives. Many of their so-called “innovations” are nothing more than existing concepts cleverly repackaged.

Four days ago, Mountain View-based Google announced AI-powered accessibility features for users with disabilities to mark Global Accessibility Awareness Day (GAAD). The move received tons of media coverage within hours of release.

On similar lines, two days prior to Google, Apple introduced a bundle of new features, including a new personal voice feature to create “a synthesised voice that sounds like them” to talk with friends or family members. The tools are specifically designed for individuals who are non-speaking or at the risk of losing their ability to speak.

Notably, in January 2023, Microsoft unveiled similar features through VALL-E, a text-to-speech AI model that can mimic a person’s voice using a three-second audio sample. The model was able to not only generate realistic audio but also preserved the speaker’s emotional tone.

Search Overhaul

At the latest Google I/O, the Sundar Pichai-led company, unveiled its revamped search interface. The updates were heralded as a major leap forward in user experience. The sleek design and enhanced functionality were indeed impressive.

However, the changes were not as groundbreaking as they initially appeared. Other search engines had already incorporated similar features and design elements. Prior to Google, You.com had built search based on generative AI in December, several months before the others did.Moreover, Perplexity.ai, a well-established name amongst the AI-powered search startups has been using AI chatbots to scourge the web’s content better than Google’s crawlers and ranking system.

Though some of these promising companies like Neeva.com have not been able to keep up with the big tech’s supremacy. Last week, the company’s founders published a blog, stating how difficult it is to build a search engine. Acknowledging Google’s predominance, they wrote, “Throughout this journey, we’ve discovered that it is one thing to build a search engine, and an entirely different thing to convince regular users of the need to switch to a better choice. One can say Google’s search “innovation” was more of an adaptation, rather than a breakthrough.”

The “Metaverse” Hype

In 2021, Mark Zuckerberg-run Facebook rebranded its entire identity as its focus shifted from social network to the company’s virtual-reality vision. The company sketched its plans to build the “Metaverse” – a digital world built over our own.

The concept captivated the public’s imagination but it wasn’t the first company to do so. The company’s marketing efforts and rebranding have generated significant hype, but the underlying concept is not entirely original.

In 2003, Linden Lab launched Second Life and businesses and universities started building their islands in Linden’s proprietary world. Two decades ago, computers were not as fast so the pioneers had a lot of limitations. In 2006, there was even a Metaverse Roadmap Summit where the Second Life spaces were explored.

Furthermore, Unity, a billion dollars gaming software company, also allowed users to have a shared “mixed reality” experience of playing a board game together, using Facebook’s Oculus 2 headsets. Earlier that year, the company behind Fortnite, Epic Games, also raised $2 billion to spend on its Metaverse plans.

Enjoying the publicity, the term ‘Metaverse’ was one of the top three contenders for Oxford Word of the Year and naturally, companies started hopping on the bandwagon. Microsoft introduced ‘Microsoft Mesh’ and NVIDIA too started building a digital twin of its own. Within two years, Metverse’s hype cycle has come crashing to the ground as the industry is going through a paradigm shift in generative AI.

The Bottom Line

While tech giants continue to dominate the industry with new updates every week, it is crucial to critically evaluate their claims of innovation. Not every update or product release deserves the praise it often receives. Behind the glitzy presentations and extensive media coverage lie repurposed ideas.

The post Big Tech Sells Old Wine In New Bottles appeared first on Analytics India Magazine.