AI Researchers are the Architects of their Own Destruction

AI Researchers are the Architects of their Own Destruction

To be scared of something you’ve built is probably the worst of all fears. It comes along with guilt and resentment of what you have done. This is the case with a lot of pioneering AI researchers such as Geoffrey Hinton and Yoshua Bengio, who are talking about abandoning their lives’ work and warn about the dangers of AI.

Bengio told the BBC that he feels “lost” over his life’s work. He is the second “godfather” of AI to come up with fears and ask for regulation of AI. Hinton was the first one to leave Google to warn the world about the problems of the pace of development of such AI technology. He claims that AI is “an existential risk”.

A protester outside a London event at which Sam Altman spoke | Getty Images

Overselling fear

Last week, hundreds of AI scientists and researchers joined in to warn that the future of humanity is at risk. The Statement on AI Risk, signed by Hinton and Bengio, also included OpenAI’s Sam Altman, Google DeepMind’s Demis Hassabis, and Stability AI’s Emad Mostaque, among others.

“Mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war,” reads the letter. Essentially, most of the people concerned about developing technology are concerned about what happens if the technology falls into the hands of “bad actors” which includes the military and terrorist organisations.

When the experts themselves come out to point out against the risks of the technology, it might be tempting and too easy to fall into the fears.

On the flip side, the people who are blooming with the AI revolution are now voting to increase developments in AI. That is also probably because making AI models like ChatGPT has now become easier with open source models. Take for example the recent debate about the need for uncensored LLMs that are outperforming bigger models.

This can also be one of the reasons why the AI experts are now becoming AI doomers. They are increasingly getting scared of letting the technology be in the hands of people who are not so well versed with the possibility of it going rogue.

But on the contrary, the third godfather of AI, Yann LeCun’s reaction to AI researchers prophesying the doomsday scenario is face palming. LeCun, the Meta AI chief, has been quite vocal when criticising people who are overestimating the capabilities of the developing generative AI models such as ChatGPT and Bard.

A statistician's view of AGI doomer drama. https://t.co/nx1pGaNVt9

— Yann LeCun (@ylecun) June 4, 2023

Another AI expert to call out the AI experts who have become AI doomers is Kyunghyun Cho, who is regarded as one of the pioneers of neural machine translation which led to the development of Transformers by Google. Cho said in a media interview that glorifying “hero scientists” is like taking people’s words as gospel without rationalising the meaning behind them.

He expressed frustration towards the developing AI discourse and how the U.S. Senate hearing was disappointing as it did not talk about the benefits of AI, but only the problems related to it. “I’m disappointed by a lot of this discussion about existential risk; now they even call it literal ‘extinction’. It’s sucking the air out of the room,” he said.

There is no agreement

The point is, for a person who has never ever deployed a single ML model in their life, it is very easy for them to become an AI doomer. They are often influenced by the hundreds of movies made about machines taking over the world. But now, if the so-called AI experts join the same narrative, with the same explanation, it seems like it is merely fear mongering, and not an actual reason.

Reminder: most AI researchers think the notion of AI ending human civilization is baloney.

— Pedro Domingos (@pmddomingos) May 3, 2023

Take for example Elon Musk, the longest running AI doomer hasn’t stopped yet. In a recent interview with Wall Street Journal, Musk said, “I don’t think AI is going to try to destroy all Humanity but it might put us under strict controls.” He still thinks there is a “non-zero chance” of AI going full Terminator on humanity.

Earlier, the petition to pause giant AI experiments and not train models beyond GPT-4 was welcomed by a lot of AI researchers including Musk and Steve Wozniak. This new petition seems a lot similar to that one, only this time it includes Altman. This suggests a lot of possibilities other than actual fear of technology taking over humanity.

Altman had asked the government to regulate AI development. At the same time, he also holds the opinion that OpenAI should be evaluated differently. This definitely questions the company’s motivation behind enforcing government regulations on AI as the regulation is a clear conflict of interest for Altman.

We all know that fear sells. “Unfortunately, the sensational stories read more. The idea is that either AI is going to kill us all or AI is going to cure everything — both of those are incorrect,” explained Cho.

The post AI Researchers are the Architects of their Own Destruction appeared first on Analytics India Magazine.

The impact of generative AI on software team productivity is… complicated

chatgpt on keyboard

Despite the potential for vast productivity gains from generative AI tools such as ChatGPT or GitHub Copilot, will technology professionals' jobs actually grow more complicated?

People can now pump out code on demand in an abundance of languages, from Java to Python. along with helpful recommendations. Already, 95% of developers in a recent survey from Sourcegraph report they use Copilot, ChatGPT, and other gen AI tools this way.

Also: How to use ChatGPT to write code

But auto-generating new code only addresses part of the problem in enterprises that already maintain unwieldy codebases, and require high levels of cohesion, accountability, and security.

For starters, security and quality assurance tasks associated with software jobs aren't going to go away anytime soon. "For programmers and software engineers, ChatGPT and other large language models help create code in almost any language," says Andy Thurai, analyst with Constellation Research, before talking about security concerns.

Also: Why your ChatGPT conversations may not be as secure as you think

"However, most of the code that is generated is security-vulnerable and might not pass enterprise-grade code. So, while AI can help accelerate coding, care should be taken to analyze the code, find vulnerabilities, and fix it, which would take away some of the productivity increase that AI vendors tout about."

Then there's code sprawl. An analogy to the rollout of generate AI in coding is the introduction of cloud computing, which seemed to simplify application acquisition when first rolled out, and now means a tangle of services to be managed.

Also: I'm using ChatGPT to help me fix code faster, but at what cost?

The relative ease of generating code via AI will contribute to an ever-expanding codebase — what the Sourcegraph survey authors refer to as "Big Code". A majority of the 500 developers in the survey are concerned about managing all this new code, along with code sprawl, and its contribution to technical debt. Even before generative AI, close to eight in 10 say their codebase grew five times over the last three years, and a similar number struggle with understanding existing code generated by others.

So, the productivity prospects for generative AI in programming are a mixed bag. For IT teams at the maintenance and support end of software, AI likely helps more than complicates, Thurai points out. "AI can also impact incident responders, site reliability engineers, and support personnel," he says.

"In their case, they can use AI to find out any precedence, how it was fixed, whether it can be automated so it won't happen again, and help with automating some of the mundane fixes to avoid constant alerting and wasting many hours in fixing things that are rudimentary. For customer service folks, it can help personalize service for individuals based on their needs, problems they faced, and the impact that was created."

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

Faster delivery of code also brings greater expectations from the business for applications that adapt more readily to changing requirements. "We are evolving toward a modeling-based approach and away from coding based on if-then-else rules," says Preeti Lobo, practice director for business integration and automation at Apps Associates .

"Today, applications need to be more intuitive and built with the individual user in mind versus a generic experience for all. Generative AI is already enabling this level of personalization, and most of the coding in the future will be developed by AI."

Still, humans are required at key points in the loop to assure quality and business alignment. "Traditional developers will be relied upon to curate the training data that AI models use and will examine any discrepancies or anomalies," Lobo adds.

The rise of AI-assisted code development also requires that technology managers and professionals assume more expansive roles within the business side. "IT pros can be expected to wear a number of new hats," says Lobo.

This growth in responsibility includes roles such as "ethical AI trainer, machine language engineer, data scientist, AI strategist and consultant, and quality assurance," she says. In addition, technology professionals will need to engage in "creating AI strategic roadmaps, as well as identifying anomalies in data structures and results."

Also: I used ChatGPT to write the same routine in 12 top programming languages

Generative AI also puts natural language processing (NLP) front and center, Lobo says. "Professionals should aim to master programming languages like Python, Java, and C++, and learn more about libraries and frameworks such as NumPy, Keras, TensorFlow, Matplotlib, and Seaborn," she says.

"But they should also look to hone strong analytical, problem solving, and critical thinking skills, as well as linguistics. Skills such as these can help exponentially in the world of NLP, a foundational factor when working with AI."

Another additional role technology professionals are assuming is coaching and supporting more people in developing and deploying their own apps. "In the past, the art of possible was limited because of the technical limitations or by the limitations of the IT departments," says Thurai.

"Now, the sky is the limit. Anyone can figure out a way to improve either the top or bottom line of any business, which can be implemented using AI to improve the business as they imagined in a faster pace than you imagined."

Artificial Intelligence

Master Gen AI For Free From Google’s Courses

Generative AI has become the talk of the town, garnering widespread attention from venture investors who are investing heavily in gen AI startups, even without the guarantee of immediate returns. By 2030, experts anticipate the generative AI market to reach an impressive $109.3 billion, signifying a promising outlook that is captivating investors across the board.

At Google I/O 2023, Google, one of the leading players in this space, introduced a suite of generative AI features for Gmail, unveiled the advanced PaLM 2 language model, showcased Med-PaLM 2 for medical applications, and highlighted the capabilities of Bard for developers. Google also announced gen AI enhancements for Google Cloud, including Duet AI, and introduced foundation models like Codey, Imagen, and Chirp.

Recently, Google also introduced nine free training courses on generative AI that also come with a completion badge.

Read more: Generative AI is Having An Edison Moment

Introduction to Generative AI

This short course of 45 minutes provides an introduction to Generative AI, its applications, and its distinctions from conventional machine learning approaches. It also includes information on Google tools that can assist in the creation of your own generative AI applications.

Introduction to Large Language Models

In this course, you will get an overview of large language models (LLMs), their definitions and explaining their potential applications. It also delves into the concept of prompt engineering, which can improve the performance of LLMs. Additionally, the module introduces various Google tools that can assist in the development of personalized Gen AI applications.

Attention Mechanism

This module focuses on attention mechanisms in deep learning and its applications in enhancing the effectiveness of different ML tasks. It explores how attention can be used to improve machine translation, text summarisation, and question answering, among other tasks. Pre-existing knowledge of ML, deep learning, NLP, computer vision python programming is necessary.

Transformer Models and BERT Model

This course introduces the key elements of the Transformer architecture, including the self-attention mechanism, and its application in constructing the BERT model. Additionally, you will understand various tasks that BERT can be employed for, including text classification, question answering, and natural language inference. Prior knowledge in intermediate ML, word embeddings, attention mechanisms, and proficiency in Python and TensorFlow are recommended.

Introduction to Image Generation

Adding to the list of interesting courses, this course is designed to teach you about diffusion models that are responsible for generating images. Throughout the course, you will learn the principles behind diffusion models, as well as how to train and utilise them on Vertex AI. Prior knowledge in ML, deep learning, convolutional neural networks (CNNs), and programming in Python is required to fully benefit from this course.

Create Image Captioning Models

This course will prepare you to build an image captioning model through deep learning. It covers the key elements of such a model, including the encoder and decoder, as well as the training and evaluation processes. Upon completion of this module, you will possess the skills to develop your own image captioning models and utilize them to generate captions for images. Prior familiarity with ML, deep learning, NLP, computer vision, and Python programming is advantageous.

Encoder-Decoder Architecture

In this course, you will understand encoder-decoder architecture, a widely used and effective ML framework for tasks involving sequences, such as translating languages, summarising text, and answering questions. The module covers the key elements of the encoder-decoder architecture and provides instructions on training and deploying these models. To make the most of this module, it is essential to have a strong foundation in Python and familiarity with Tensorflow.

Introduction to Responsible AI

Building products with a focus on ethical AI is important. So this course by Google where you will understand the significance of responsible AI, as well as how Google incorporates responsible AI into its products is a need of the hour.

Introduction to Generative AI Studio

Google introduced a new Gen AI Studio at Google I/O. So this course gives an introduction to Generative AI Studio, a tool within Vertex AI and how to use it effectively. Generative AI Studio allows you to create and personalise generative AI models, enabling you to incorporate their abilities into your own applications.

Generative AI Explorer (Vertex AI)

The Generative AI Explorer – Vertex AI Quest is a series of labs on using Generative AI on Google Cloud. It covers the Vertex AI PaLM API family, including models like text-bison, chat-bison, and text embedding-gecko. You’ll learn about prompt design, best practices, and applications like text classification and summarisation. The quest consists of four modules: foundation models, model tuning, PaLM API, and Text Embedding API.

Read more: AI Cloud Wars: Azure AI vs Vertex AI

The post Master Gen AI For Free From Google’s Courses appeared first on Analytics India Magazine.

AI Unlikely to Replace Humans in VFX, Says Red Chillies’ COO

India is one of the largest contributors of VFX in international movies. Now with the rise of generative AI tools and platforms such as Wonder Studio, Kapnetix, WithPoly, NVIDIA’s Omniverse and Magic3D this industry is at the verge of disruption, and there is a fear around job loss to AI, or automation.

Companies like Red Chillies, Digital Domain, Phantom FX, Prana Studios, MPC, and Prime Focus that have worked on major Hollywood movies, including the likes Deadpool, X-Men Apocalypse, and Spider-man: Homecoming, are under grave theat.

“Although AI is still in its early stages and has limitations, it is revolutionising the VFX industry by enhancing efficiency and delivering more realistic outputs. While AI is unlikely to replace humans in the VFX sector, it empowers VFX artists to work more effectively,” said Keitan Yadav, chief operations officers of Red Chillies Entertainment—one of India’s leading vfx makers. He said that the industry is eagerly exploring the potential of AI and other emerging technologies to elevate the quality of VFX and filmmaking outputs.

However, this brings us to something that concerns the vfx ecosystem in India. The important question here is—whether this wave of advancements in tech and tools lead to Hollywood movie makers reducing the reliance on Indian VFX companies, and would they start leveraging AI-powered tools to produce it themselves. To this, Yadav said that not everything in the industry can be automated due to its nature.

Further, he said that global studios have already established their presence in India, attracted by the existing film industry and the readily available workforce. Initially, India focused on low-level tasks but has now advanced to more complex work for both the Indian and Hollywood markets.

“Certain jobs may speed up and certain jobs may not anymore. Be required to be sent to India, but I’m sure there are a lot of other kind of VFX jobs, which will eventually keep coming to India because of the Reasons I mentioned earlier. And since it cannot be completely automated ever, that is always be. The human intervention. Which is going to be very critical. so, That’s what my take is. I think it’s gonna be the golden decade of VFX”

Yadav believes that the VFX industry will continue to grow in India due to its own ecosystem, cost-effectiveness, and human intervention being crucial. He predicts a promising future for the animation and AVGC (Animation, Visual Effects, Gaming, and Comics) workspace in India, driven by government support, evolving manpower resources, and cost advantages.

Yadav began his career as a VFX artist in 1996. He forayed into Film VFX in 2003, as a VFX Supervisor and VFX Producer, starting with the film “Main Hoon Na,” produced by Shah Rukh Khan’s company, Red Chillies Entertainment. Red Chillies Entertainment later established a visual effects studio called redchillies.vfx in 2006 and Yadav has been at the helm ever since.

Generative AI Limitations & Opportunities

Yadav said while certain tasks can be done quickly and easily using generative AI tools, there is always a demand for new and never-before-seen work. He said that each project presents a new challenge, highlighting the needs for a professional setup and level in order to consistently deliver high-quality animation and VFX.

Citing Wonder Studio, he said that promising tools are being developed, but they are still in their early stages and not considered fully professional. Yadav believes that as the industry evolves, these tools will become more advanced and accessible for VFX studios to incorporate into their pipelines, providing numerous opportunities.

Yadav believes that as the industry evolves, these tools will become more advanced and accessible for VFX studios to incorporate into their pipelines, providing numerous opportunities.

He acknowledged that while AI may not provide perfect results initially, it has the potential to automate many processes currently done manually. He emphasised that the industry has already embraced the use of machine learning and AI as tools, and over time, with research and development, more tasks will be taken up by AI-powered software. Furthermore, he notes that the software and tools used in the VFX industry are standardised globally, and any advancements in AI technology will impact studios worldwide, including those in India.

Inside Red Chillies Studio

“See, the VFX industry has always been at the forefront in terms of using technology and the latest machines. The fastest machines, the best tools and best software. It’s a very forward industry and …… it (AI/ML) has already started, impacting the VFX world. so there are some pipeline softwares…which have already started incorporating machine learning and AI To help the big projects, automate, a lot of the stuff, which previously used to get done manually,” avered Yadav.

“So there are some softwares which have already come out. And they are also using AI… maybe not fully but to an extent and I’m sure in the future, this use will keep growing,” he added.

At Red Chillies, Yadav said that AI is being used to create, alter, or enhance human faces, replacing the conventional methods of 3D modeling and sculpting with machine learning algorithms that analyse real images of actors to render realistic models. Additionally, AI tools also enable facial substitution and face swapping, as demonstrated in the short film “Face Swap,” showcasing the impact of deepfakes.

Citing Disney Research, Keitan said that there are a lot of beauty software that can age and de-age actors based on their images from different time periods. He acknowledged that while AI may not provide perfect results initially, it has the potential to automate many time-consuming tasks.

“On the artistic side. Yes. There are some beauty software, which basically age and deage an actor, okay? So they take all the images of the actors, right? From his 18 years old till the current time and requireingly. You know, deage the person, those tools are also almost out,” said Yadav.

“So these are all uses of AI in our industry,” he added, saying that the use of AI will surely automate a lot of things. “Of course, they may not give you 100% results, but say initially it may start giving you from 30% and then they will improve it to 50, 60, 70, and so on. And maybe one day, they may be able to give us almost near perfect results. But that is gonna take time.”

Yadav said as research and development progresses, more tasks will be taken over by AI-powered software. He noted that the VFX industry in India uses standardised software and hardware that is also used globally, and any advancements in AI technology will impact studios worldwide, including those in India.

The post AI Unlikely to Replace Humans in VFX, Says Red Chillies’ COO appeared first on Analytics India Magazine.

6 Possible Use Cases for Falcon 40B

Falcon 40B, developed by the Technology Innovation Institute (TII) in Abu Dhabi, is a large language model (LLM) with 40 billion parameters, making it one of the largest LLMs globally. It is an open-source model, freely available for research and commercial purposes.

Although still in development, Falcon 40B shows great potential for various applications. TII encourages researchers and developers to explore its use cases and share their findings with the community. In exciting news, Falcon 40B is now royalty-free for both commercial and research use, meeting the demand for inclusive access to AI worldwide.

It has achieved the top rank on Hugging Face‘s leaderboard for large language models, surpassing competitors like Meta’s LLaMA and Stability AI’s StableLM. The software is licensed under Apache 2.0, ensuring end-users have access to any patents covered by the software, establishing security, availability, and an effective governance model.

Falcon 40B offers a wide range of potential use cases.

Natural language processing (NLP): Falcon 40B, a remarkable language model, offers a wide range of applications in the field of natural language processing (NLP). It can effectively handle tasks like text classification, sentiment analysis, and question answering. For instance, it can categorise text into various types like news articles, blog posts, or product reviews. It can also evaluate the sentiment of text, determining whether it is positive, negative, or neutral. Furthermore, Falcon 40B can provide informative answers about a given text, including details about the author, the main topic, or the key points.

Generative text tasks: In the realm of generative text tasks, Falcon 40B shines by generating creative content in different formats such as poems, code, scripts, musical pieces, emails, and letters. It can create rhyming poems with specific meter, generate syntactically correct and meaningful code, and produce well-structured scripts with realistic dialogue.

Chatbots: Chatbot development is another area where Falcon 40B excels. It enables the creation of chatbots capable of engaging in natural conversations with users, ranging from answering customer queries about products or services to providing customer support.

Machine translation: Machine translation is made easier with Falcon 40B, as it can proficiently translate text from one language to another, facilitating the communication of ideas and information across linguistic boundaries.

Data augmentation: Falcon 40B also proves valuable in data augmentation, as it can generate synthetic data resembling real-world data, which enhances the performance of machine learning models trained on the augmented dataset.
Research: Researchers can leverage Falcon 40B to explore new avenues in the utilisation of large language models (LLMs) for a multitude of tasks. It offers possibilities for advancing NLP techniques, machine translation methods, chatbot development, as well as novel applications like creative writing and game development.

The post 6 Possible Use Cases for Falcon 40B appeared first on Analytics India Magazine.

What to Expect at Apple WWDC 2023

Apple WWDC 2023

After Google and Microsoft talking about AI and Copilot at their conferences, Apple is ready for its Worldwide Developers Conference (WWDC). To be held on June 5th, there are a lot of expectations from the company to dive into the space of generative AI, something that they have been shying away from. Apart from that, there are a dozen announcements that Apple is expected to definitely make.

A new era begins. Join us for #WWDC23 on June 5 at 10 a.m. PT.
Tap the ❤ and we’ll send you a reminder on event day. pic.twitter.com/T1pDvEzvys

— Apple (@Apple) May 31, 2023

Mixed Reality Headset

Apple’s long-anticipated venture into mixed reality hardware is set to make waves at WWDC. With an emphasis on a blend of real and virtual worlds, this standalone headset, rumored to be called Reality Pro, promises to outshine its competitors. Meta just released its own VR headset ‘Quest 3’, just ahead of Apple’s event.

Boasting a 4K resolution per eye, six cameras for full body motion tracking, and an M2 chip, this headset will likely offer an immersive experience. The highlight, however, lies in the innovative software that is expected to introduce hand gestures, gaze detection, and voice commands through Siri for navigation.

Object Capture and RealityKit enable the integration of real-world objects seamlessly into the AR experience. In fact, during WWDC 2022, Apple showcased a glimpse of this capability. The introduction of ARKit 6 has already unlocked a range of exciting possibilities, including 4K video, 3D object capture, face tracking, motion capture, and scene geometry.

Given Apple’s commitment to AR, it is highly anticipated that WWDC 2023 will bring further enhancements and additions to the ARKit framework, expanding the potential of AR even further.

iPod coming back?

Just a month back, Apple had filed a patent for a technology to bring back one of its most popular and loved products – the iPod. The patent includes for music, videos, and books, with the additional feature of wireless earbuds. Though there has been no confirmed news from the company, there were reports in 2020, that the company is working on a top-secret iPod.

Privacy and healthcare to be possibly a focus

Privacy has clearly been the selling point of Apple all these years. Recently, the company has been promoting its privacy priority through various advertisements, specifically focusing on the Apple Watch.

One of the company’s most notable features in this space is the electrocardiogram (ECG) feature. Integrating that with the Apple Health App and Watch, while also protecting the data might be the next big step for the company. “The Health app is built to keep data secure and protect your privacy. Your data is encrypted and you are always in control of your health information,” is written in the launch of the Health app.

Generative AI, Apple chatbot, update to Siri

Something that everyone has been talking about – generative AI. Apple has been known to be committed to its privacy, and that is what probably makes it not dive into the field of large language models and chatbots. Most recently, the company decided to ban use of ChatGPT for its employees. It would be interesting to see if the company plans to develop something for its internal use.

On the other hand, Apple’s love for federated learning has always been evident in everything it does. With its emphasis on privacy and building for edge use cases, the company can release something similar, or better than ChatGPT that can run on single devices. The chances look bleak, but are surely possible.

Moreover, Apple recently unveiled its own Transformers architecture, specifically designed for their Apple Silicon technology. This innovative architecture supports LLM and GPT models, which have been crucial to Apple’s advancements in generative AI.

Apple has taken significant steps in this direction by launching secure and private LLM models that operate directly on the device. As a result, it is highly likely that Apple will integrate generative AI capabilities to enhance Siri’s performance and offer users an even more powerful and interactive virtual assistant experience.

Apple Maps

An update to Apple Maps has been long awaited. The time is probably here as the company has been showing improvements in its offering by introducing a lot of features that are actually better than Google Maps. The new ‘Look Around’ perspective changes while moving the map, giving it a more live-like feeling, covering a lot more details.

Read: Can Apple Draw Better Maps than Google?

Revolutionary 15-inch MacBook Air

Apple enthusiasts can look forward to a surprising addition to the MacBook Air lineup at WWDC. Speculations suggest a 15-inch version, previously only available in the MacBook Pro series, could soon be within reach. Powered by the M2 chip and featuring an improved display and larger battery, this MacBook Air will provide a new level of portability and performance.

iOS 17 and iPadOS 17: More Freedom for Users

The upcoming iOS and iPadOS updates are rumored to introduce a significant change: sideloading. In compliance with European Union competition rules, users may soon be able to install apps outside the App Store.

This opens up possibilities for direct downloads from the web and third-party stores, providing a more diverse app ecosystem. Additionally, these updates are expected to bring improvements such as a revamped Control Center, enhanced health and journaling features, and a tablet-native Health app for iPads.

Apple Watch’s Biggest Software Update: watchOS 10

WatchOS 10 is anticipated to bring a significant redesign to the Apple Watch. With a focus on widgets reminiscent of the original Glances feature, the interface will offer quick access to information. While apps will still be present, the emphasis will be on providing snippets of data for users to glance at conveniently. Apple might introduce this new interface as an option initially, allowing users to ease into the revamped experience.

A Sneak Peek at macOS 14

Details about macOS 14, the next major update for Mac computers, remain scarce. However, it is expected to continue Apple’s tradition of incorporating features from its iOS counterpart. Mac users can look forward to the unveiling of macOS 14 at WWDC, with the official release likely scheduled for October.

The Wildcard: The M3 Chip

While not confirmed, rumors of the M3 system-on-chip debut at WWDC persist. This next-generation chip, boasting enhanced performance and efficiency, could potentially accompany the anticipated 15-inch MacBook Air. Additionally, Apple might surprise us with a refreshed 13-inch MacBook Air and an M3-powered iMac later in the year. However, don’t hold your breath for Pro models, as they are not expected until 2024.

The post What to Expect at Apple WWDC 2023 appeared first on Analytics India Magazine.

Is Adobe Firefly Killing Midjourney, DALL.E?

Adobe has been a popular choice for designers for ages. However, two years ago, when OpenAI brought generative AI into the design world with DALL.E, it took the world by storm.

On the other hand, when Adobe introduced its Firefly generative AI service in March, it received a pretty lukewarm response as designers felt that it was not worth the hype. But just last week, Adobe surprised everyone by outshining all the generative AI design startups like DALL.E, Midjourney, and Stable Diffusion with its groundbreaking ‘Generative Fill’ feature in Photoshop.

Read more: Design Wars: Microsoft Designer vs Canva

Generative Fill: Boon for Adobe

Built on NVIDIA Picasso, Firefly’s generative fill is a game-changer. It enhances the design process by transforming interactive elements into photorealistic images and enabling the creation of fresh styles.

The Generative Fill tool removes unwanted objects from photos while preserving their aesthetic appeal. It also generates detailed artwork from text prompts and allows customisation of vectors, brushes, and textures. Apart from this, Adobe Firefly also revamps content creation for social media and marketing, generating distinctive posters, banners, and social media posts from text prompts.

The new 'Generative Fill' AI in Photoshop is mind-blowing.
Finally, an AI feature for practical use,
Priority #1, let's fix the Pope's photo: pic.twitter.com/HcDFjcli3r

— Lorenzo Green 〰 (@mrgreen) May 25, 2023

Adobe is excelling in generative AI for several reasons. Additionally, Adobe stands out with its great user interface (UI), exemplified by Adobe Firefly, which offers an intuitive design and helpful prompts for users unfamiliar with AI. A twitter user pointed out that her Adobe’s UI is so smooth and easy that even her 73 year old artist friend could figure it out in one go.

Furthermore, Adobe’s popular software, Photoshop, integrates advanced generative AI features like inpainting and outpainting, similar to DALL.E and Stable Diffusion.

Leading in a Packed Market?

Firefly isn’t alone in the gen AI design application market. Midjourney, Stable Diffusion, and DALL.E have been leading players in the market. But looks like, it is changing.

Canva introduced an AI image generator using Stable Diffusion, while Microsoft Designer uses OpenAI’s DALL.E 2 with a method called unCLIP.

While Microsoft Designer has targeted consumers initially, it may soon expand to the enterprise sector. It uses DALL.E 2, which is trained on stock images, making it suitable for corporate use.

Stable Diffusion, on the other hand, is fully open source, while DALL.E 2 and Midjourney are not. Due to this, Canva’s usage of Stable Diffusion excels in detailed artistic illustrations but struggles with complex prompts and generic images like logos. To make things worse, it has also generated inappropriate content. Additionally, amid the GPT-4 craze, OpenAI has forgotten about DALL.E as there have been no recent updates on it. However, the choice between DALL.E 2 and Stable Diffusion depends on user needs.

Nevertheless, Adobe brings the best of both worlds with Firefly. Adobe’s decision to develop Adobe Firefly as a separate free product has also proven strategic, allowing users to try generative AI without a subscription and enticing them to become subscribers after experiencing the product.

Lastly, Adobe’s collaborations with Google and NVIDIA further strengthen their position. Partnering with Google ensures safe AI development while leveraging NVIDIA’s expertise in training models enhances the quality of generated images. Through its focus on commercial safety, transparency, user-friendly interfaces, integrated AI features, free offerings, and strategic collaborations, Adobe has emerged as a frontrunner in the field of generative AI.

Read more: Lessons from YouTube for Gen AI Copyright Mess

Adobe is Winning the Copyright Game

AI copyright is the current buzzword. The need to regulate AI has sparked a global debate. Italy banned ChatGPT for privacy concerns, while the EU is introducing the AI Act. The US released a blueprint for an AI Bill of Rights, and OpenAI CEO Sam Altman called for a new government agency to regulate large AI models.

However, artists have been protesting against the same for a while now. AI-generated content has faced opposition from Shutterstock, Getty Images, as well as artists who see it as a threat to their intellectual property. Adobe has also been very vocal about the same as echoed by Adobe’s chief product officer, Scott Belsky, “We have never, ever used anything in our storage to train a generative AI model.”

Adobe is clearly winning the copyright game by training its generative AI model on a dataset based on Adobe Stock, along with openly licensed work and public domain content where the copyright has expired. Additionally, Adobe emphasises transparency by actively participating in the Content Authenticity Initiative, which aims to make content more transparent and trustworthy.

Firefly has also incorporated metadata into the images that it generates. This metadata includes information about the source of the image, the AI model that was used to generate the image, and can be used to identify AI-generated artwork and to protect the rights of the creators.

The incorporation of generative AI in Adobe builds on its existing capabilities in addition to upholding copyright integrity, positioning it as a huge threat to alternative image generators.

Read more: Stuck in Traffic? Blame it on BMTC

The post Is Adobe Firefly Killing Midjourney, DALL.E? appeared first on Analytics India Magazine.

Google is Now a Classified Ad page! 

Ranking on Google has always been an extremely difficult task for new websites, as the top positions are typically occupied by sponsored posts. Google’s main source of revenue comes from advertising, so it’s not surprising that they are exploring ways to profit from their AI-powered search engines. The search engine is already considering generating revenue through AI-written responses to search queries.

For example, when a user asks about factors to consider when buying a CPU, the generative AI not only answers the query but also suggests CPUs along with marketplace ratings. It wouldn’t be costly for a company like Google to earn money from marketplaces through affiliate marketing, as tech blogs have been doing this for years.

However, there is one issue: the articles generated by the AI-based search engine are of lower quality and merely imitations of the originals.

Currently available in limited beta testing, Google’s new Search Generative Experience (SGE) transforms the site from being a search engine that provides links to the best content into a publication that offers its own mini-articles. Instead of hiring expert writers for this task, Google employs an AI that analyzes data from human-authored content and produces advice that lacks expertise or authority.

However, as tech experts have pointed out, the AI-generated write-up is not the best one to follow. When following the generic write-up, the search engine presents a list of outdated processors that are no longer among the best CPUs available today. The top choice is a Ryzen 7 5800X3D, which hasn’t been the top processor for a year, and there’s also a link to a Core i5-10400, which is three generations old. This is not helpful advice.

Similarly, when a user searches for “hiking backpacks for kids” on the new version of Google, the AI could respond with an ad for a particular backpack, along with tips for choosing the best backpack for a child, according to Google.

Jerry Dischler, vice president and general manager of ads, mentioned in an interview that this new approach to interacting with Google search is simpler and more helpful. He believes it will lead to new commercial opportunities and deliver a delightful user experience.

Paid is new Organic

Amazon is known as a marketplace where merchants can come and sell their products. However, as soon as a product starts performing well on the platform, Amazon replicates it and sells it under the brand name ‘Amazon Basics.’

Google is now planning to follow a similar path. It used to be recognized as one of the best search engines, delivering accurate results with incredible speed. However, Google has now launched its own generative AI-based search engine, where users can find Google’s AI-written content at the top of the result page, followed by sponsored and organic posts.

Before Microsoft introduced the idea of a generative AI-based chatbot producing search results, even Google believed that expertise should be the sole metric for ranking. Just a few months ago, the search engine revealed that it would use E-A-T (Expertise, Authority, Trust) to determine content rankings. In other words, content from experienced authors with subject matter expertise should rise to the top, while unreliable advice from content farmers without a proven track record should fall to the bottom.

However, it seems that Google has found a way to generate extra revenue from the new generative AI-backed search engine. It is suspected that in the future, when concerns are raised about the AI picking up illegitimate sources to generate answers (as it is doing currently), Google may offer a solution where websites pay Google when their website is selected to generate an answer.

The post Google is Now a Classified Ad page! appeared first on Analytics India Magazine.

6 Reasons Why We Won’t Get GPT-5 Anytime Soon

OpenAI has been criticised for its opaque decision-making process. Unperturbed, CEO Sam Altman has graced the world with some valuable insights. In an interview with Humanloop, the oft-criticised celebrity CEO spoke about the future of OpenAI. While this interview shed some light on some of OpenAI’s most profitable products, it also reflected a culture of overpromising and under-delivering.

While the interview itself presents a few surface-level points, it also delves into some of the scaling issues that OpenAI has been facing. Along with this, some of Altman’s statements in this interview directly conflict with his statements in the past, creating more opacity for ‘OpenAI’. However, one thing is for certain, GPT-5 isn’t coming anytime soon, not even in 2024.

Here are 6 reasons why we won’t get GPT-5 any time in the near future.

Not enough GPUs

In the interview, Altman delved into the scaling issues OpenAI is facing. According to him, OpenAI is ‘extremely GPU-limited’, which has rippled into delays for their short-term plans. This means that plans like longer context lengths (more than 32,000 tokens) and fine-tuning for specific dataset will be delayed.

Due to the shortage, even the basic functionality of the API has been compromised. According to the interview, OpenAI is offering a dedicated capacity offering, which will offer faster and more dependable API access. However, to access this, customers have to commit to a $100,000 spent on the platform.

A bigger focus on GPT-4

While GPT-4 was one of the biggest upgrades for OpenAI, it is clear that Altman isn’t done with this model. The first and biggest item on the roadmap was a cheaper and faster GPT-4. By improving its flagship model, OpenAI wishes to reduce the ‘cost of intelligence’ as far as possible.

This means that the company will likely spend time and resources to optimise GPT-4 even further instead of focusing on GPT-5. In addition to this optimisation, OpenAI is looking to increase the context window to up to 1 million tokens. Other upgrades on the roadmap include fine-tuning APIs and a stateful API for GPT-4.

Restricted by scaling laws

Altman has been vocal in the past about how the size of LLMs won’t matter as much moving forward. However, it seems that OpenAI’s researchers have changed his mind. According to the interview, its internal data has found that scaling laws continue to hold well.

This means that GPT-5, when it releases, will not be as big of a leap as GPT-2 to GPT-3.5. This is because models will likely ‘double or triple’ in size, as opposed to changes in the order of magnitude.

Promised features first

Multimodality was one of the biggest new features of GPT-4. Demos of the model being capable of generating a whole website from just a sketch set the Internet on fire, but it seems even ChatGPT Plus users have to wait for multimodality.

In the interview, Altman stated that multimodality will only come in 2024, blaming the lack of GPUs. Until multimodality comes to GPT-4, GPT-5 is still a far-off dream.

ChatGPT is the last product

Altman also stated that ‘OpenAI would not release products beyond ChatGPT’. This is mainly because OpenAI wants to be safer for competing with their customers with other offerings. He also stated that there was a history of platform companies having a killer app.

This solidifies OpenAI’s future goals as a platform company rather than a product company. As time goes on, their platforms will only get more capable. This can also be interpreted to mean that GPT-5 might only get more delayed in favour of better platforms.

Regulatory risks

Sam Altman has also been very vocal when it comes to AI regulation. He is also one of the biggest signatories for the new AI safety letter which advocates for more stringent regulation of AI, citing existential risks.

While Altman said that he continues to believe in the importance of open source, he also defended OpenAI not releasing their models to the public. This is mainly due to his skepticality towards the large number of companies that have the capabilities to host and serve LLMs.

With the host of problems OpenAI is facing to scale up to the success of ChatGPT, it seems that other projects might be put on hold. While even the GPT-4 API has not been updated for the past 2 months, GPT-5 is still a long-distant dream until OpenAI gets it together.

The post 6 Reasons Why We Won’t Get GPT-5 Anytime Soon appeared first on Analytics India Magazine.

Will Peter Deng Help OpenAI Recover from ChatGPT Hangover?

OpenAI announced that it has onboarded Peter Deng as the VP of consumer product, where he will be leading the company’s consumer product efforts, alongside driving design and engineering teams behind ChatGPT with the goal of making AI accessible to all.

“We’re at the crossroads of a new era in computing, and there is so much to be built,” said Deng. He believes that we have barely scratched the surface in building productivity and creative tools that delight users.

Prior to joining OpenAI, Deng was the chief product officer at cloud collaboration platform Airtable. He was responsible for leading the company’s product strategy and execution, where he helped the company scale significantly, alongside raising funding at a $11 billion valuation.

Deng has worked for 18 years as a product lead and manager, in most of the major tech companies like Google, Instagram, Facebook (now Meta), Uber to name a few.

After graduating from Stanford with a Master’s in Arts (MA) in media studies and communications, he began his career at Google in 2006. This was a period of large changes; Google went public in 2004, alongside its acquisition of Youtube (2006) and DoubleClick (2007), and Deng speaks of gaining valuable insights into product development and strategy.

ChatGPT Hangover

It was surely a wild time to be working at Google. But, it is even crazier now, given all the hype around OpenAI, and the sky-high expectations from the consumers.

When ChatGPT was launched in November last year, it literally broke the AI hype cycle, where within three months, the chatbot scaled to 10 million users within three months. Nothing like seen before in the history of product adoption. In contrast, Instagram took almost a year to reach the same milestone.

“ChatGPT surprised us a little bit,” said Altman, adding that he thought it would have been the 10th fastest growing product in history, and not the number one.

“Someone said to me over the weekend, you shipped an AGI and I somehow, like, am just going about my daily life and I’m not that impressed,” quipped Sam Altman, in an interview with Lex Fridman, with reference to how users are not impressed post-ChatGPT era.

Clearly, OpenAI is struggling to create its next big breakthrough, or a product miracle once again, thanks to ChatGPT’s wild success. That explains why the company hired Deng to get them out of the ChatGPT hangover.

Builder Deng Creates Magic

Deng’s contribution to consumer products has been impeccable, and has shaped the company he’s previously worked in to scale products to the next level. “I have always found myself to be a bit of a productivity geek, and a builder,” said Deng.

In 2007, Deng joined Facebook at a nascent stage. He played a vital role in various products, including Chat, Events, Groups, and the iconic News Feed. His contributions spanned the redesign of the News Feed for real-time updates, the introduction of structured data collection for user profiles, and the transformation of Groups into vibrant collaborative spaces. Deng’s work solidified Facebook’s position as a leading social network, connecting billions of people worldwide.

Deng’s ability to navigate diverse product landscapes led him to assume pivotal roles at other tech giants. In 2013, he took on the position of Head of Product at Instagram when the platform was still relatively small. He led the team at Instagram and drove exponential growth, reaching hundreds of millions of users.

Aggressively consumer centric, Deng joined Uber In 2017 as head of Rider. Then present in around 700 cities, he focused on solving local problems instead of a blanket approach. For example, Uber launched Uber Lite, a simpler version of the app in India, saving storage and data, and estimated 1.5 million new users because of it. His focus shifted to the complexities of a real-world, two-sided marketplace. Leading consumer product efforts, Deng navigated the intricate dynamics between drivers and consumers, steering Uber through critical product transformations, and contributing to the company’s successful IPO.

A recurring pattern in his career is his work with relatively new companies and introducing game-changing features and products to it.

The Subtle Art of Mindfulness

Spending all day in meetings (only the essential ones), as a product manager Deng mentions how it is important to be living in the moment.

“Everyday I try to take a little bit of time to sit and be present with the fact that i’m planning for the future, and ironically that makes me more present then, and it makes me more present during that day,” said Deng, sharing his learning around applied mindfulness, which helps people become more compassionate and collaborate better.

The post Will Peter Deng Help OpenAI Recover from ChatGPT Hangover? appeared first on Analytics India Magazine.