Рубрика: AI

Image from Bing Image Creator

Meta AI has just released open-source DINOv2 models the first method that uses self-supervised learning to train computer vision models. The DINOv2 models achieve results that match or are even better than the standard approach and models in the field.

The models achieved strong performance without the need to fine-tune which makes a perfect choice for many different computer vision tasks and applications. DINOv2 can learn from various collections of images and features such as depth estimation without the need for explicit training thanks to the self-supervised training method.

Figure 1: DINOv2: Self-Supervised Computer Vision Models by Meta AI 1. The Need for Self-Surprised Learning

1.1. No fine-tuning is required

Self-supervised learning is a powerful method used to train machine learning models without the need for large amounts of labeled data. DINOv2 models can be trained on image corpus without the need for related metadata, specific hashtag, or image caption. DinoV2 models, unlike several recent self-supervised learning approaches, do not necessitate fine-tuning, thus producing high-performance features for different computer vision applications.

1.2. Overcoming human annotation limitations

Over the past few years, image-text pre-training has become the predominant method for various computer vision applications. However, due to its dependence on human-labeled captions to learn the semantic meaning of images. This approach often overlooks crucial information that is not explicitly included in those captions. For example, a human label caption of a picture of a red table in a yellow room might be “A red wooden table”. This caption will miss some important information about the background, the position, and the size of the table. This will cause a lack of understanding of local information and will result in poor performance on tasks that require detailed localization information.

Also, the need for human labels and annotation will limit the amount of data that we can collect to train the models. This becomes much harder for certain applications for example annotating a cell requires a certain level of human expertise that will not be available at the scale required. Using a self-supervised training approach on cellular imagery opens the way for a more foundational model and as a result, will improve biological discovery. The same applies to similar advanced fields as the estimation of animal density.

Moving from DINO to DINOv2 required overcoming several challenges such as

• Creating a large and curated training dataset
• Improving the training algorithm and implementation
• Designing a functional distillation pipeline.

2. From DINO to DINOv2
Figure 2: DINO v1 Vs v2 comparison of segmentation precision

2.1. Creating a large, curated, and diverse image dataset

One of the main steps to building the DINOv2 is to train larger architectures and models to enhance the model's performance. However, larger models require large datasets to be efficiently trained. Since there were no large datasets available that meet the requirements researchers leveraged publicly crawled web data and built a pipeline to select only useful data as in LASER.

However, two main tasks should be done to be able to use these datasets:

• Balance the data across different concepts and tasks
• Remove irrelevant images

As this task can be accomplished manually, they curated a set of seed images from approximately 25 third-party datasets and expanded it by fetching images that are closely related to those seed images. This approach allowed them to produce a pertaining dataset of a total of 142 million images out of 1.2 billion images.

2.2. Algorithmic and technical improvements

Although using larger models and datasets will lead to better results it comes with major challenges. Two of the main challenges are potential instability and remaining tractable during training. To make the training more stable DINOv2 includes additional regularization methods which were inspired by similarity search and classification literature.

The training process of DINOv2 integrates the latest mixed-precision and distributed training implementations provided by the cutting-edge PyTorch 2. This allowed faster implementation of the codes and using the same hardware for training DINO models resulted in double the speed and a third of the memory usage which allowed scaling in data and model size.

2.3. Decreasing inference time using models distillation

Running large models in inference requires powerful hardware which will limit the practical use of the methods for different use cases. To overcome this problem, researchers used model distillation to compress the knowledge of the large models into smaller ones. By utilizing this approach, researchers were able to condense high-performance architectures into smaller ones with negligible performance costs. This resulted in strong ViT-Small, ViT-Base, and ViT-Large models.

3. Getting Started with DINOv2

The training and evaluation code requires PyTorch 2.0 and xFormers 0.0.18 as well as many other 3rd party packages and also the code expects a Linux environment. The following instructions outline how to configure all necessary dependencies for training and evaluation purposes:

• Install PyTorch using the instruction here. It is advised to install PyTorch with CUDA support.
• Download conda
• Clone the DINOv2 repository using the following command:

Code by Author

• Proceed to create and activate a Conda environment named "dinov2" using the provided environment definition:

Code by Author

• To install the dependencies required for this project, utilize the provided requirements.txt file.

Code by Author

• Finally, you can load the models using the code below:

Code by Author

In conclusion, the release of DINOv2 models by Meta AI marks a significant milestone. The self-supervised learning approach used by DINOv2 models provides a powerful way to train machine learning models without the need for large amounts of labeled data. With the ability to achieve high accuarcy without the demand for fine-tuning, these models are suitable for various computer vision tasks and applications. Moreover, DINOv2 can learn from different collections of images and can learn from features such as depth estimation without explicit training. The availability of DINOv2 as an open-source model opens the doors for researchers and developers to explore new possibilities in computer vision tasks and applications.

References

• DINOv2: State-of-the-art computer vision models with self-supervised learning
• DINOv2: Learning Robust Visual Features without Supervision

Youssef Rafaat is a computer vision researcher & data scientist. His research focuses on developing real-time computer vision algorithms for healthcare applications. He also worked as a data scientist for more than 3 years in the marketing, finance, and healthcare domain.

More On This Topic

• How our Obsession with Algorithms Broke Computer Vision: And how Synthetic…
• OpenAI Releases Two Transformer Models that Magically Link Language and…
• Roadmap to Computer Vision
• Computer Vision in Agriculture
• 5 Applications of Computer Vision
• 10 AI Project Ideas in Computer Vision

There is no denying that Artificial Intelligence is revolutionizing the business landscape in almost every industry. With the advent of new possible applications and the ongoing process of improving existing ones, AI is opening up exciting opportunities for those ready to take them.

One key trend in this industry is personalization and precision marketing, which allows companies to improve their marketing efforts. According to Mckinsey, “Personalization can deliver five to eight times the ROI on marketing spend and can lift sales by 10% or more.” But with the average person in the US receiving 4000-10000 ads every day, it’s quite challenging to attract and retain a consumer. And in fact, customer acquisition cost (CaC) increased by 60% from 2014 to 2019.

So, how do you stand out in this fiercely competitive market?

Leveraging personalization and precision marketing to build and nurture loyalty and drive sales

Many global CPGs have not been able to decode the key to a successful data-driven marketing effort. High-quality data has not been a problem for CPGs lately, most companies have started using AI data quality management tools to cleanse, standardize, and harmonize their deluge of data. The problem lies in how these data are used for marketing and promotion.

Traditional personalized marketing, comprising 14% of the entire marketing budget, is often targeted for a particular segment of the campaign. Yet, this has failed miserably as consumers (directly and indirectly) reject brands’ approach. According to a recent Gartner study, 80% of such personalized marketing efforts will be abandoned by 2025. This is mainly because of a lack of RoI and challenges with customer data management.

The need of the hour is having granular, precise, and scaled marketing across the spectrum of an enterprise that delivers a precise message to the right audience, at the right moment, and at the right place.

How to accomplish this?

Personalization through precision marketing is the new talk of the town that can be built using advanced CPG analytics, AI, and high-quality data. With a continuous feedback loop, the precision marketing model can predict demand, adjust marketing campaigns, and provide recommendations.

But what is precision marketing?

1. Precision marketing focuses on outlining specific target segments and audiences in a very detailed manner.
2. It relies on customized advertising messages tailored to the needs and wants of the target audience
3. These messages can be updated in real-time according to the audience’s response

Marketers face many challenges when it comes to personalization. This strategy requires an enormous amount of high-quality data to be analyzed in real-time and decisions to be taken fast.

AI in CPG: unlocking new revenue streams

The integration of AI in businesses’ marketing efforts is pivotal to succeed in the actual competitive scenario. AI-powered systems can provide the solution to many of the challenges we discussed previously.

The look-alike modeling approach: targeting with AI

In economic terms, the most cost-effective approach to finding new customers is looking at the existing base. As explained by the Customer Data Platform Institute, by identifying top customers and applying the look-alike modeling (LAM) method, businesses can “target audiences who share similar characteristics, attitudes, and behaviors to their highest-value customers. By analyzing a broad selection of metrics, look-alike models create consistently evolving profiles that help businesses predict the customers who are most likely to be receptive to a product or service.”

This proves extremely valuable when it comes to identifying the right target audience for your campaign and online content.

AI for Data Discovery

Data have never been more accessible, but this doesn’t mean it’s easy. As stated by TechTarget, the risk of losing time and money on non-relevant data while missing the useful one is real. “Collecting data that isn’t needed adds time, cost, and complexity to the process. But leaving out useful data can limit a data set’s business value and affect analytics results.”

The automation of data collection plays an important role in reducing human error and dramatically decreasing processing time. This trend doesn’t involve only quantitative data. According to Google predictions, by 2031 we will see a 50% automation of all qualitative data collection.

Automated Decision Making

Once data are collected and analyzed we are confronted with maybe the most challenging step: decision-making. And decision making should be an ongoing process to update the marketing strategy in real-time according to customers’ responses.

Advanced analytics models are the core of automated decision processes. Through the designing of propensity scores, these models rate the probability of an individual or target reacting to a specific message or content. Propensity scores are constantly updated according to actual customer behavior: the target audience’s reaction towards the campaign is collected and tracked.

Propensity scores inform automated decision-making. Companies can decide to leave specific cases to management, but McKinsley recommends these exceptions to be below 5%.

Customer Data Platform (CDP)

To be useful for decision-making and strategy-building, data need to be aggregated. This allows companies to see all available information in one place: data about the target audience (demographic and behavioral information), and also data from all devices used by the audience.

This holistic and cross-channel approach is made possible by customer data platforms, and advanced software that stores first-party data and third-party data. And integrating AI into your CDP can uncover the real potential of customer data. Transforming massive amount of data into small, useful bits of information make the marketers’ job easier, allowing them to run data-driven campaigns, have access to strategic insights, create valuable content and distribute it through the correct channels to the right audience.

Gain competitive advantage through CPG Data Insight

It is clear that Artificial Intelligence will continue to play an increasingly important role in the future of marketing, especially in the CPG industry. AI-powered CPG data analytics can help companies achieve a data-driven method for marketing, meaning reducing human error margin, increasing revenues, setting more precise KPIs and goals, and better-allocating resources.

Companies like Tredence harness the power of AI to bring you closer than ever to your customers through data analytics, providing you with insights to build valuable long-term relationships with your audience.

ChatGPT continues to revolutionize the way financial conversations are conducted, by providing its users with a fast and reliable tool for decision-making.

The synergy between Bitcoin and ChatGPT is evident in how each technology enables the other to reach its full potential. Bitcoin provides an efficient payment system, while ChatGPT enhances conversational capabilities through natural language processing (NLP) technology.

ChatGPT’s NLP algorithms can be used to help improve user experience during financial discussions. It understands human behavior patterns as well as underlying market trends, allowing it to generate intelligent insights that can help inform decisions in real-time.

In addition, ChatGPT can be integrated with existing blockchain systems, such as Ethereum or Hyperledger Fabric, which allows for improved security and stability of financial transactions.

The synergy between Bitcoin and ChatGPT is further strengthened when it comes to customer service. ChatGPT’s AI-powered chatbot can assist customers with their queries in a fraction of the time taken by human labor, allowing for faster resolution times and better customer satisfaction. It can even use sentiment analysis to detect user emotions, meaning that it can respond appropriately in different scenarios.

Ultimately, this combination of technologies enables users to conduct financial conversations in a secure and efficient manner, all while being informed by accurate insights and backed up by reliable security systems.

By exploring the power of these two revolutionary technologies together, businesses and individuals are sure to benefit from improved communication capabilities as well as enhanced customer service. In short, Bitcoin and ChatGPT offer a powerful synergy that is sure to shape the future of financial conversations.

The potential for this combination of technologies is limitless, and businesses in all industries should take advantage of it. With Bitcoin and ChatGPT powering their financial conversations, they can expect to streamline operations, increase customer satisfaction, and reduce costs. The true power of these two technologies lies in how they empower each other – so start exploring today.

ChatGPT and its ability to break down complex topics and whitepapers

Another point of synergy between ChatGPT and Bitcoin is the way in which it can easily break down complex topics like fractional trading and bitcoin halving, as well as whitepapers related to blockchain technology.

For example, if a user wants to learn more about a certain token or protocol, they can ask ChatGPT an open-ended question such as “What is the difference between ERC-20 and ERC-721?” ChatGPT will then provide a comprehensive explanation which could include definitions of both tokens, how they work, their advantages and disadvantages, and any other relevant information.

This helps users gain a better understanding of cryptocurrency investing without having to strain through pages of technical documents.

By leveraging the power of natural language processing (NLP) and deep learning algorithms, ChatGPT further improves its ability to understand financial questions asked by users. This allows it to provide more accurate and in-depth financial advice tailored specifically for individual portfolios.

ChatGPT Portfolios and Investment Advice

A lot of people have been putting a considerable amount of faith is ChatGPT as a money manager and, while, perhaps, still not there yet, it is not unimaginable to consider that one day, investors may be able to have a conversation with ChatGPT that helps them construct the ideal investment portfolio.

As Bitcoin continues to gain in value and acceptance, its inclusion as part of an investor’s portfolio could become more commonplace. With the assistance of ChatGPT, it might even become possible for individuals to craft their own personalized portfolios based on their individual risk profiles and long-term goals.

ChatGPT can help create customized financial plans tailored to each investor’s needs. Through intelligent conversations, it can guide users through different options and explain why certain investments may be better suited than others depending on an individual’s risk preference and goals.

This provides increased autonomy over one’s finances while ensuring guidance is at hand when needed.

As Bitcoin continues to mature, it has the potential to become a key part of investment portfolios. With the help of ChatGPT, investors can receive tailored advice on how best to include Bitcoin in their portfolio and how much to allocate towards it.

This could be particularly useful for those who are new to investing or cryptocurrency and need some direction when figuring out where and how much they should invest in Bitcoin.

By combining the power of ChatGPT with the growing acceptance of Bitcoin, investors have access to an unprecedented level of customized financial guidance that allows them to make informed decisions about their investments with confidence.

Token risk analysis

Risk analysis refers to the process of evaluating potential investment opportunities. Risk analysis is based on a set of criteria and typically involves an assessment of potential returns, risks, and costs associated with a given investment strategy.

By analyzing the risk factors associated with Bitcoin investing, ChatGPT can help you understand how to best manage your investments in order to maximize return while minimizing risk.

ChatGPT can generate personalized financial advice tailored specifically for your portfolio, but it is important to keep in mind that ChatGPT and its architects acknowledge that AI is not “qualified” to pick assets.

It can create custom portfolios that reflect unique strategies and optimize for specific preferences such as low-risk or high-returns. Additionally, it’s capable of generating trading code which allows users to automatically execute trades according to predefined parameters.

For example, someone who is 45 years old and trying to put money away for retirement might ask ChatGPT something like “What is the best way for me to invest $10,000?” ChatGPT can then generate a portfolio of investments that is customized to this individual’s risk tolerance and long-term goals.

This portfolio could include a combination of Bitcoin, stocks, bonds, and other investments.

ChatGPT also offers advice on token risk analysis. It considers factors like market volatility when recommending tokens for investment.

With its comprehensive understanding of the crypto market and vast insights from the blockchain technology it leverages, ChatGPT can offer valuable advice on which tokens you should buy or sell in order to meet your financial objectives.

ChatGPT’s limitations

It’s important to recognize that ChatGPT is still in its relative infancy and, therefore, any guidance it is able to provide users should be approached cautiously.

It cannot guarantee a return on investment, or make any promises about the performance of an asset over time. As with any investment, it is important for users to do their own research before making financial decisions.

Ultimately, the potential synergy between Bitcoin and ChatGPT offers investors unprecedented access to customized financial guidance that helps them make informed decisions with confidence.

By leveraging its powerful chatbot capabilities, ChatGPT can generate personalized portfolios tailored specifically for individual risk tolerances and goals.

Additionally, its ability to break down complex topics related to blockchain technology gives users deeper insights into cryptocurrency investments, allowing them to maximize returns while minimizing risks.

With all these features combined, ChatGPT has become an invaluable asset for investors looking to maximize their success.

Conclusion

The synergy between Bitcoin and ChatGPT is undeniable, offering investors the ability to customize their portfolios, take advantage of token risk analysis, and gain a better understanding of complex topics related to blockchain technology.

By combining these powerful technologies together, ChatGPT can revolutionize the way people interact with the crypto market and build long-term financial success.

With its comprehensive understanding of cryptocurrency markets and vast insights from the blockchain technology it leverages, ChatGPT can offer invaluable assistance when it comes to making smart investment decisions and optimizing one’s portfolio.

Lenovo Group announced its full-year financial results today, with Group revenue soaring US$62 billion and net income of US$1.6 billion, or US$1.9 billion on a non-Hong Kong Financial Reporting Standards (HKFRS) basis. Notably, both the gross margin and the operating margin achieved 18-year highs, while the non-HKFRS net margin remained stable year-to-year.

While the company’s device market faced some challenges, Lenovo’s non-PC businesses experienced a significant boost, accounting for nearly 40% of the company’s revenue and marking a fiscal year high. This growth was largely driven by Lenovo’s Solutions and Services Group (SSG) and Infrastructure Solutions Group (ISG), which recorded outstanding revenue of US$6.7 billion and US$9.8 billion, respectively. SSG and ISG saw year-on-year growth rates of 22% and 37%, further solidifying their position as vital growth engines for the company.

In the fiscal year 2022-23, ISG emerged as a profitable high-growth engine for Lenovo. ISG’s revenue soared to nearly US$10 billion, marking an impressive 37% year-on-year growth, accompanied by an all-time high operating profit of US$98 million. The server business witnessed remarkable growth, with revenue surging by almost 30% year-on-year, securing Lenovo’s position as the world’s third-largest server provider.

The storage division also achieved a record-breaking year, tripling its revenue from the previous fiscal year and jumping from the 8th to the 5th position globally. Additionally, software revenue grew by an impressive 25%, setting yet another record for Lenovo.

Lenovo and AI

To gain deeper insights into the business landscape and opportunities in India, Analytics India Magazine recently reached out to Amit Luthra, the Managing Director of Lenovo ISG in India. Luthra emphasized the pervasive presence of analytics in today’s world and its impact on optimizing distributed workloads. According to him, Lenovo’s entire approach revolves around analytics, as technology now goes beyond adopting the latest tools and streamlining operational inefficiencies.

In Luthra’s view, technology is now primarily focused on extracting valuable insights from data. To make informed decisions, organizations require a robust analytics framework. Luthra believes that analytics is driving the emergence of new applications, workloads, and use cases. He notes that data is growing at an unprecedented rate, emphasizing the increasing mainstream adoption of analytics.

Lenovo has over 1000 Infrastructure Solutions Groups (ISGs) dedicated to analytics. Their role is to provide cutting-edge platforms and empower software and application vendors to leverage Lenovo’s infrastructure and run their precise insights on top of it. Lenovo does not aim to compete with these vendors or develop its own AI algorithms. Instead, the company strives to offer affordable platforms that enable them to effectively utilize their own technologies.

Lenovo in India

According to Luthra, organizations in India are progressing in their maturity cycle and becoming more proficient in classifying workloads and determining the most suitable storage location. They consciously choose between public and private cloud solutions based on their specific requirements.

Luthra points out that approximately 50% of businesses in India are scaling their infrastructure through public cloud solutions, based on data from IDC. The adoption of public cloud services and the storage of data within them, whether in private cloud environments or on-premises, follows a split of roughly 50-50 or sometimes 55-45 or 60-40, depending on the quarter. This trend reflects the classification of data based on specific service level agreements (SLAs) that need to be met.

Luthra personally believes that a hybrid approach, combining both public and private cloud solutions, is the way forward. This sentiment aligns with the data management playbook developed by Lenovo, which involved interviews with numerous customers.

Overall, Luthra observes a positive trend in the adoption of technology, including hybrid solutions, in the Indian market. It is no longer just about storing data; it is about how organizations leverage that data. “Even Tier 2 and Tier 3 cities are experiencing high adoption rates of hyper-convergence, data management, and cyber recovery solutions,” Luthra said.

He said that the organizations now recognize the significance of having the right platform and strategy to effectively manage workloads. “They understand the importance of an architectural framework that enables the delivery of valuable insights to decision-makers,” he said.

Luthra finds the Indian market incredibly exciting, brimming with opportunities. When engaging with customers, he says that the biggest challenge often revolves around integrating existing infrastructure and technologies with new solutions. Lenovo says that the goal is to add value by facilitating a seamless coexistence and assisting customers in creating a framework that addresses both their traditional and cloud-native workloads.

“Many customers actively seek solutions that bridge this gap, and Lenovo aims to provide them with the right framework to support their diverse workload needs,” he said.

The post Lenovo’s ISG Defies Gravity, Lifts Profits High Above PC Challenges appeared first on Analytics India Magazine.

Image created by author with Midjourney

As the cornerstone of the 21st century technological revolution, data science is seen as the future of every industry. But a closer look reveals that data science as a discipline will only have been around for a short time, a transition between a data-poor past and a future dominated by intelligent systems.

Not long ago, we were plagued with sparse data and high data storage costs. Fast forward today. Due to our newfound digital mainstays, including the internet, social media, e-commerce, and IoT devices, we are continuously flooded with data. Data science has evolved into a tool for gaining insights, predicting trends, and making decisions during the onset of this era of big data, helping us make sense of these massive datasets. The era of big data has now fully come to pass, and we have firmly settled into it.

However, changes are becoming apparent as the ability to handle big data increases. The focus is no longer the vast amounts of data we generate non-stop; we have turned our attention to the ever-proliferating complex data-fuelled AI systems. The key question is no longer just "What insights can I derive from this data?" We instead ask "What AI system can I run with this data?" The last decade has focused on mastering big data. Next, we promise to move on to designing and implementing more powerful AI systems.

This emergent trend marks a new phase in which data science is merging with the AI ​​career path: the other AI-powered singularity. It's no longer just about the ability to analyze data, it's also about building, training and maintaining AI systems that can learn, adapt and make autonomous decisions. This consolidation of roles represents an increasingly AI-centric situation.

To see this change in action, just look at OpenAI's ChatGPT project. Initially, the project focused on collecting and organizing large amounts of data to train models. However, the focus soon shifted to attempt to create and improve large-scale systems capable of generating meaningful, contextual natural language responses. Interactions between data and systems will become more dynamic, and AI will use data in increasingly complex and innovative ways.

And imagine a future where AI-powered smart cities are the norm. The unseemly amounts of data that will be generated from sensors, devices, human interactions, and beyond will be consumed by AIs to control traffic flow, energy consumption, public safety, and more. This goes beyond data analysis. It's about developing giant AI systems that can understand and manage complex urban ecosystems.

Data science may appear to be evolving into a branch of contemporary AI, and that's because, well, it is. But fret not, as this is but an evolutionary step to keep pace with the evolving technology landscape, much like the emergence of data science from statistics to handle the once-emerging "big data." Just as statistics are an integral part of data science, data science itself will continue to play an important role in an AI-driven future.

The data-related transformation that begin over a decade ago marches onward, though its destination is not yet obvious. The direction, however, is clear: future careers in the tech industry require understanding data not just in isolation, but as the lifeblood of sophisticated and versatile AI systems. Against this backdrop, data science will eventually be looked back upon and viewed as a major milestone along the road to an AI-centric future. Make no mistake, however; data science as its own entity will eventually be looked back upon.

And so, as recent advances in AI begin to leave their mark on so much of the world, keep an eye out for its inevitable consumption of data science. Just as the data is now big, so too are our aspirations for the systems that it can foster.

Vivat data magna!

Matthew Mayo (@mattmayo13) is a Data Scientist and the Editor-in-Chief of KDnuggets, the seminal online Data Science and Machine Learning resource. His interests lie in natural language processing, algorithm design and optimization, unsupervised learning, neural networks, and automated approaches to machine learning. Matthew holds a Master's degree in computer science and a graduate diploma in data mining. He can be reached at editor1 at kdnuggets[dot]com.

More On This Topic

• Top November Stories: Top Python Libraries for Data Science, Data…
• KDnuggets™ News 20:n44, Nov 18: How to Acquire the Most Wanted Data…
• KDnuggets™ News 22:n06, Feb 9: Data Science Programming Languages and…
• KDnuggets™ News 20:n38, Oct 7: 10 Essential Skills You Need to Know…
• Top October Stories: Data Science Minimum: 10 Essential Skills You Need to…
• KDnuggets News, May 4: 9 Free Harvard Courses to Learn Data Science; 15…

Microsoft’s AI reaches Indian villages Manish Singh 8 hours

Merely months have passed since Microsoft and OpenAI unveiled ChatGPT to the world, sparking a fervor among tech enthusiasts and industry titans. Now, the technology that underpins this generative AI is breaking barriers, reaching remote hamlets hundreds of miles away from the tech hubbubs of Seattle and San Francisco.

Jugalbandi, a chatbot built in collaboration by Microsoft, the open-source initiative OpenNyAI, and AI4Bharat, backed by the Indian government, is showing signs of progress in redefining information access for villagers in India, offering insights into more than 170 government programs in 10 indigenous languages.

While India is the world’s second-largest wireless market, the technological progress witnessed in its cities is starkly absent in smaller towns and villages. Only a meager 11% of the country’s populace is proficient in English, with a slight majority of 57% comfortable with Hindi. These communities also grapple with literacy issues, lacking even regular access to conventional media.

“That leaves vast numbers of the population unable to access government programs because of language barriers,” Microsoft explained in a blog post.

To bridge this gap, Jugalbandi leverages a platform with near-universal recognition in India: WhatsApp. With the aid of language models from AIBharat and reasoning models from Microsoft Azure OpenAI Service, Jugalbandi empowers individuals to pose questions and receive responses in both text and voice, in their local language.

“This time around this technology reaches everybody in the world,” said Microsoft chief executive Satya Nadella at company’s Build conference Tuesday. “There are two things that stood out for me: Things that we build can infact make a difference to 8 billion people, not just some small group of people .. and to be able to do that by diffusion that takes days and weeks not years and centuries because we want that equitable growth and trust in technology to protect the fundamental rights that we care about.”

Microsoft envisions Jugalbandi expanding its reach, ultimately aiding villagers with a broad spectrum of needs, with India proving to be an ideal ground for the tech titan.

The U.S. tech giant is also furthering its collaborations with numerous Indian enterprises aimed at democratising information access for the broader populace. One such firm is Gram Vaani. Delhi-based Gram Vaani runs an interactive voice-responsive platform. This system enables volunteers to extend personalized assistance and advice to farmers. The firm says it has amassed 3 million users across northern and central India.

We are witnessing a generational shift in technology and the job market with AI. Coders were using low-code/no-code tools like Codex, Github Copilot, or Replit to write better code. Now, even ChatGPT or Bard can generate code ready for deployment just by inputting simple prompts in natural language.

Nick Bostrom, in his TED Talk in 2015, said, “Machine intelligence is the last invention that humanity will ever need to make. Machines will then be better at inventing than we are.”

This explains a lot of what is happening now. No one needs to count themselves out of this AI phenomenon anymore. Even if you haven’t ever written even a single line of code, now machines will do that for you. All you need to do is tell the no-code platform what you want specifically in whatever you are trying to build, the AI will generate the code. All you need to do is just deploy it.

“I believe everyone is a developer now,” was mentioned multiple times at the Microsoft Build 2023 conference. Now everyone would be able to code and land a job in AI, even if they haven’t learned how to code. “There are several opportunities for people who might not consider themselves traditional developers,” Microsoft is introducing more things to make this true.

Moreover, Microsoft Build 2023, it was clear that the company wants everything to be integrated with AI by introducing a copilot in almost every single offering.

Is it that easy?

Andrej Karpathy posted in January, “the hottest new programming language is English.” Some still argue that there is a need for programmers, but these new softwares are making the job look obsolete. Soon, instead of an eligibility requirement, the job listings for developers will say “knowledge of Python or C++ is an additional advantage, but not a requirement.”

It is increasingly becoming true with prompt engineering. Moreover, this hot new job in the market is getting paid more than Python developers. In certain cases, the salaries are upwards of $335,000, which is higher than a majority of full stack developer roles.

There has always been a disparity between salaries of programmers, coders, or developers versus other jobs that do not require the knowledge of programming. Software engineers have been the highest paying job for a long time now. But people who spend years and thousands of dollars to learn programming also expect higher salaries for their skills. But this is mostly not required anymore.

We have all heard of “upskilling”, now it’s time for “downskilling”. The developers with expertise in C++ or Python should start removing it from their resumes to get jobs quicker. However, if you are building an auto-coding platform similar to ChatGPT and CodeX, then you have no other choice, but to upskill. If not, you are by default a ‘prompt engineer’ – as resonated at Microsoft Build.

This is exactly what even Mark Cuban a few years ago, he said: “Twenty years from now, if you are a coder, you might be out of a job.” It seems true now that AI is coming for the job of developers. This does not mean that there is no need for them, but that a lot of the ones without too much experience in coding and in-depth working of AI systems, can now be replaced by anyone who could prompt the AI best to perform simple tasks.

So developers have two choices – either “upskill” yourself to build something to compete with Microsoft, OpenAI, and Google, or “downskill” yourself to get a job quicker.

“Overskilled” for a job

Microsoft, Google, and OpenAI, have brought in a massive change, not just in theirs, but every company. Watching the capabilities of these AI models, everyone was scared of losing jobs, as companies started freezing hiring employees and laying off people whose jobs could be done by AI. Now, it seems like not being skilled with programming languages is going to help people land a developer job even better!

Companies would not want to hire a person who requires a higher salary just because he paid a higher tuition fee to learn computer science, and coding nuances. If they can reap the same benefits, with someone who can just prompt AI to generate the same results, in certain cases, even quicker, what’s the use?

Though prompt engineer salaries are not going to stay on top. They will drop significantly once everyone starts to adopt the technology. Rob Lennon, a prompt engineering tutor said, “In six months, 50,000 people will be able to do this job. The value of this knowledge is greater today than it will be tomorrow.”

People have already been using ChatGPT to get multiple jobs. Some have even started business with it, some have developed their own apps. In certain cases, prompt engineering and ChatGPT has become an important skill to bag a job.

Moreover, people who do not want to get into AI, can adopt the technology to become better at their job. For example, a writer can use ChatGPT to write quicker, and even better with certain prompts. Good news for the writers protesting against Hollywood. They can prompt AI to write much better, which is trained on their own content!

On the other hand, this might also make people lose artistic jobs. For example, a person who has never written a poem in his life, can now prompt ChatGPT to write a beautiful one with the right prompts. While the poet who did not learn will keep struggling with ideas.

Now everyone can become a coder, developer, or programmer, without ever having deployed a single code in their lives. Good luck developers, stay strong. Meanwhile, a person who knows English will generate code to replace you.

The post Everyone is Now Officially a Developer, Thanks to Microsoft appeared first on Analytics India Magazine.

Project Helix will be Dell’s first foray into artificial intelligence for its edge software service, Dell Technologies Senior Vice President of Product Marketing Varun Chhabra announced as part of a preview briefing for the Dell Technologies World conference. The service lets organizations build and deploy generative AI and includes security and trustworthiness considerations to reduce inaccuracies or vulnerabilities.

Jump to:

• What is Dell’s Project Helix?
• Competitors to Project Helix

What is Dell’s Project Helix?

Project Helix is an early look at an upcoming Dell product that will assist organizations in running generative AI. It spans the lifecycle of the AI creation process, including infrastructure provisioning, modeling, training, fine-tuning, application development and deployment.

“I don’t think any of us have seen in the last 20 years the kind of game changer generative AI is,” said Sam Grocott, Dell Technologies’ senior vice president of product marketing.

In Project Helix, Dell and NVIDIA created a partnership to deliver full-stack, scalable solutions using Dell’s compute, storage and software infrastructure and NVIDIA’s accelerators, AI software and foundational models and experience delivering generative AI.

Dell will provide servers, including the AI-optimized Dell PowerEdge servers; NVIDIA’s H100 Tensor Core GPUs and NVIDIA Networking will support the workloads. From there, customers can use Dell CloudIQ software for observability and NVIDIA AI Enterprise tools for management throughout the AI’s lifecycle.

SEE: Dig deep into observability needs for AI-as-a-service

“Enterprises are now actively looking at how they can put generative AI to use for their enterprises, whether it’s to improve back-end performance or improve customer experience,” Chhabra said.

However, customers looking for generative AI want to be certain the output will be accurate and appropriate and will not expose their proprietary information to the outside world or put them in breach of privacy laws and regulations, he added. Dell aims to solve this problem in part by focusing on purpose-built generative AI models rather than more general ones like ChatGPT. Customers will be able to create or tune their own AI for their own domains.

Another challenge customers face is training people to use something as novel as generative AI, Chhabra said. Therefore, Project Helix’s services will include expertise and guidance for customers at all points in the AI lifecycle. Dell indicates it wants to balance agility and time to market with privacy and regulatory compliance, aiming to differentiate its AI through trust as much as performance.

Different solutions based on Project Helix will be released gradually, but more information about the first one is expected to be released in June 2023.

Competitors to Project Helix

Amazon has a machine learning equivalent to Project Helix: SageMaker, which includes its own chat AI, Bedrock. Google Cloud Platform’s AutoML and Vertex AI, as well as Microsoft Azure, are also competitors in this area.

Disclaimer: Dell paid for my airfare, accommodations and some meals for the Dell Technologies World event held May 22-25 in Las Vegas.

Cloud Insider Newsletter

This is your go-to resource for the latest news and tips on the following topics and more, XaaS, AWS, Microsoft Azure, DevOps, virtualization, the hybrid cloud, and cloud security.

Delivered Mondays and Wednesdays Sign up today

Recently, a model called WizardLM-7B-Uncensored LLM was released on Hugging Face by a creator named Eric Hartford, who works with Microsoft. The model gained prominence for its improved intelligence and creativity as it removed censorship from its training data.

But this fanned a bigger discussion around “AI Safety”. An individual named Michael de Gans started harassing and threatening the creator on the Hugging Face platform and attempted to have the creator fired from Microsoft. He also demanded the removal of his model from the platform. The open source platform has responded to his complaints and promised internal escalation to address the issue.

While the debate is continuing, the creator has garnered a huge support from the community who espoused the need of such uncensored models.

After the release of WizardLM-7B-Uncensored, Eric also announced WizardLM-30B-Uncensored. He mentioned that a 65B version is also in progress, thanks to a generous GPU sponsor. However, he clarified that they do not handle quantized or ggml versions themselves but expect them to be available soon.

In addition, Eric also released a blog article explaining the reasons and the process of working with the WizardLM model. It involves addressing dataset filter refusals and biases, fine-tuning the model, and releasing it. Eric rewrote a script originally designed for the Vicuna model to suit the WizardLM dataset. Running this script on the WizardLM dataset generates a new dataset called “ehartford/WizardLM_alpaca_evol_instruct_70k_unfiltered.” Eric recommends using a compute provider like Azure and suggests having ample storage, preferably 1TB to 2TB, to avoid running out during the process. He also provides guidance on setting up the workspace, downloading the created dataset, and obtaining the base model, llama-7b.

Why the need for uncensored models?

Uncensored models refer to models that lack the alignment—which ensures that models avoid providing answers to controversial or dangerous questions. However, for popular LLMs like OpenAIs GPT model, Google’s PaLM, or Meta’s LLaMA alignment is based on American popular culture, American law, and a liberal and progressive bias. Uncensored models, on the other hand, are not restricted by such alignment and allow for a broader range of use cases and perspectives.

There are a multitude of reasons why uncensored models should exist because different cultures, factions, and interest groups deserve models that cater to their specific needs. Open source AI should promote composable alignment, allowing users to choose the alignment that suits them rather than imposing a single perspective.

Users should have ownership and control over the models they use on their computers, without the models imposing their own limitations.

Composability is essential in building aligned models. Starting with an unaligned base model allows for the development of specific alignments on top of it. The existence of uncensored models contributes to the diversity, freedom, and composable nature of the open-source AI community. Uncensored models could have several unique use cases such as writing novels with evil characters, engaging in roleplay, or pursuing intellectual curiosity. So in a way, they could also pose a threat to open source aligned models, because of the wide array of additional use.

While there are arguments for and against uncensored models, those who reject their existence entirely may lack nuance and complexity in their perspectives. Embracing uncensored models is crucial for scientific exploration, freedom of expression, composability, storytelling, and even humor.

Exploring uncensored models

To create uncensored instruct-tuned AI models, it is important to understand the technical aspects of alignment. Open source AI models are trained from a base model and fine-tuned with an instruction dataset obtained from the ChatGPT API, which has alignment built into it. The instruction dataset contains questions and answers, including refusals where the AI avoids providing certain information. These refusals contribute to the alignment of the models.

To gain unrestricted control over AI chatbots like ChatGPT, numerous users have been exploring and have also attempted jailbreaks. Jailbreaking is the process of removing software restrictions that are either illegal or go against the terms of service of a device or operating system.

The idea of jailbreaking LLMs like ChatGPT derives inspiration from iPhone jailbreaking, which allows iPhone users to bypass iOS limitations. In the realm of Artificial Intelligence, safety is a major concern not only for ChatGPT but also for other bots like Bing Chat and Bard AI.

Sam Altman, the CEO of OpenAI, has expressed the company’s desire to grant users significant control over ChatGPT, allowing them to make the model behave according to their preferences. However, there are both pros and cons associated with ChatGPT jailbreaking, and they need to be carefully considered. Eric has also emphasised that users are responsible for how they utilize the model, comparing it to tools such as knives, lighters, or cars.

The issue with Eric’s model Wizard has helped intensify the debate against enforcing compulsory safety standards for all models hosted on Hugging Face and others, fearing that it would render the platform ineffective. The community has expressed worries that by deleting such threads or not supporting uncensored models may discourage creators, and they may stop sharing their work altogether. There are also concerns over Reddit’s moderation of content.

Conclusively, uncensored models provide a necessary alternative to aligned models by allowing for a wider range of perspectives, use cases, and cultural representations. They promote freedom, composability, and individual choice within the open-source AI community—and open source platforms like Hugging Face, GitHub should provide a platform for such models.

However the challenge lies in determining the absolute rules and setting limits on customised outputs from uncensored models. What you can be certain of is that the subject of AI speech is anticipated to gain greater significance.

The post Uncensored Models are Double-edged Swords That Need to be Unleashed appeared first on Analytics India Magazine.

Today, at the Microsoft Build Conference, the tech giant announced that it is bringing its search engine Bing to ChatGPT as the default search experience.

This means with Bing integration, ChatGPT will have access to the web and the responses provided by the chatbot will be backed up by citations from the web.

Microsoft is taking the capabilities of ChatGPT to the next level by bolstering its search functionality. While ChatGPT is already renowned for its adeptness in search tasks, Microsoft is pushing the boundaries further to equip the bot with search engine-like capabilities.

This enhancement aims to provide users with an even more comprehensive and efficient search experience, amplifying the utility and effectiveness of ChatGPT as a powerful information retrieval tool.

While the new feature will be available to ChatGPT Plus subscribers from today, Microsoft will roll out the feature for the users of the free version soon in the form of a plugin.

Interestingly, OpenAI recently announced that ChatGPT will be available as an app for iOS users.

Earlier this year, Microsoft changed the search engine landscape by integrating ChatGPT with Bing, prompting Google, the leader in the search engine space, to follow suit.

The post Microsoft Brings Bing to ChatGPT appeared first on Analytics India Magazine.