Hidden AI Daily October 16

Hey there. Welcome back to another edition of Hidden AI Daily, where we dive into the exciting world of AI and technology.

Today, we have a packed edition with Replit AI, Tata Consultancy Services, TCS, OpenAI, and more updates. We’ll also explore the use of generative AI in the Indian film industry and its impact on consent.

So, grab your favorite beverage and prepare for an episode filled with fascinating insights and discussions. Let’s jump right in.

Replit AI Empowers 23 Million Developers with AI Coding Assistance

Replit’s Mad Obsession with ‘AI for All’

Replit AI, a platform aimed at democratizing AI and coding, has made its suite of AI products accessible to over 23 million developers.

One of their key offerings is Ghostwriter, a generative AI-powered coding partner that assists developers with code completion. The company aims to make AI assistance accessible to all developers, especially those who cannot afford premium services like GitHub Copilot.

Replit’s updated LLM Replit Code V1, 5.3b is open-sourced and performs on par, if not better, than other code completion models like StarCoder and CodeLlama. To enhance productivity and enable complex software coding, Replit has partnered with Google Cloud. This partnership grants them access to Google Cloud’s infrastructure and Vertex AI.

Replit focused on the top 30 programming languages used by developers on their platform when training their LLM model. However, they faced challenges obtaining GPUs due to the ongoing chip crisis, which impacted their large-scale AI deployment.

India is an essential market for Replit, and they are addressing latency concerns by establishing local servers and engaging in localization efforts. The popularity of Replit’s mobile app among Indian students indicates its relevance and effectiveness in the Indian educational landscape.

Overall, Replit aims to provide global access to its products for learning and skill development while potentially opening doors to employment opportunities and entrepreneurship.

TCS and Microsoft Strengthen AI Partnership

TCS Seeks a Boost in AI Partnership with Microsoft

Tata Consultancy Services (TCS), one of the leading Indian IT firms, is strengthening its partnership with Microsoft to develop AI-based software services. The collaboration involves using Azure OpenAI, a cloud-based AI tool developed by Microsoft, specifically GitHub Copilot.

The goal of this partnership is to improve profit margins for TCS, and they are doing so by offering solutions such as fraud detection for financial services and personalized customer services for retailers using the AI tool.

To further enhance client relationships and leverage the expertise of senior executives, TCS’s CEO, K Krithivasan, has initiated organizational changes within the company.

This move aligns with the overall trend in the Indian IT industry, where firms are shifting their focus towards technologies like big data, machine learning, analytics, cloud computing, and AI. TCS aims to enhance its capabilities for deeper engagement with these transformations by closely aligning itself with clients’ future investments in these areas.

 In addition to their partnership with Microsoft, TCS collaborates with Google. They will offer Google’s cloud generative AI services, Vertex AI, through their Model Garden platform.

OpenAI’s Stateful API Revolutionizes Development

OpenAI Aims for a Bolder Presence in the Developer Arena

OpenAI has recently announced updates to enhance the developer experience and make creating and marketing AI-driven applications more cost-effective. One of these updates is the introduction of the Stateful API, which will retain request histories using added memory. This feature could reduce operational costs by up to 20 times.

In addition, OpenAI is launching the Vision API, enabling developers to create scalable solutions for image analysis and processing. This new offering will give developers powerful tools to develop innovative applications that leverage computer vision technology.

However, while OpenAI’s latest offerings are exciting, it’s important to note that they won’t be as groundbreaking as a GPT-4.5 or GPT-5 release. The company faces competition from tech giants like Microsoft, Google, and Amazon in the developer platform domain.

To stay competitive and relevant in the evolving AI landscape, OpenAI must balance maintaining its exclusive ecosystem and diversifying its business model. By introducing cost-efficient development tools like the Stateful API and Vision API, OpenAI is taking steps towards achieving this goal.

Low-paid Workers Label AI Data

Millions of Workers Are Training AI Models for Pennies

Today’s news explores the world of low-paid workers training AI models for pennies. According to a report from WIRED, millions of workers in countries like the Philippines and Colombia are labeling training data for AI models used by tech giants such as Amazon, Facebook, Google, and Microsoft. The global data collection and labeling market was valued at $2.22 billion in 2022 and is expected to reach $17.1 billion by 2030. One popular crowdsourcing platform called Appen hires these workers who tag training data for AI algorithms.

However, their wages are low, ranging from $0.02 to $0.50 per task. For many workers in developing countries like Venezuela, Appen is their sole source of income amidst economic hardships. They often work long hours to earn a decent income.

This has led experts to view platforms like Appen as a form of data colonialism, as workers in developing countries label images that will be used in AI models primarily benefiting the global north.

Workers on crowdworking platforms like Appen face numerous challenges as technology evolves, requiring them to update their skills constantly. Many workers feel dissatisfied with their pay and want additional compensation for time spent waiting for assignments or conducting necessary research. They also desire recognition as valued employees rather than disposable tools and hope for unionization within the industry.

Indian Court Protects Anil Kapoor’s Personality Rights

Bollywood Goes Gaga Over Generative AI

The use of generative AI and deep fakes in the Indian film industry is debated. The Delhi High Court recently safeguarded actor Anil Kapoor’s “personality rights” and ordered 16 entities to stop using his name and image without permission. This raises concerns about how generative AI may be used to create new content without the consent of celebrities.

Zee Entertainment, a major player in the Indian film industry, uses AI to enhance creativity and intuition but believes that humans will always play a role in the creative process. They see AI as a tool to augment human capabilities rather than replace them.

The Rabbit Hole, a brand film agency, has embraced generative AI tools like Dall-e and Midjourney to enhance video creation. By leveraging these tools, they can explore new storytelling and visual effects possibilities.

Biren Ghose of Technicolor Creative Studios predicts that famous personalities will have digital and physical avatars with specific licenses for commercial use. This suggests opportunities for real-life celebrities and their virtual counterparts in future media productions.

Technicolor Creative Studios has already utilized AI-based neural rendering with permission from a personality to create a Mastercard commercial. This demonstrates the potential for collaboration between AI technology and human creatives in the entertainment industry.

While Hollywood faces a SAG-AFTRA strike lasting over three months, the eastern film industry has a more collaborative and collectivist approach. Thomas, an expert in this field, believes this collaborative culture should also be brought into the AI domain for positive outcomes.

Customize AI Responses with SteerLM

NVIDIA AI Unveils SteerLM: A New Artificial Intelligence Method that Allows Users to Customize the Responses of Large Language Models (LLMs) During Inference

In the latest news from MarkTechPost, NVIDIA Research has introduced a new AI method called STEER-LM. This technique allows users to customize the responses of large language models, LLMs, during inference. STEER-LM simplifies the customization process through a four-step supervised fine-tuning process. It trains an attribute prediction model to evaluate qualities like helpfulness, humor, and creativity.

Using attribute-conditioned supervised fine-tuning, STEER-LM can generate responses based on specified attributes. STEER-LM is even more impressive because it allows users to fine-tune attributes during inference in real time. This means you can adjust and personalize the model’s responses while interacting with it.

According to NVIDIA’s tests, STEER-LM outperformed existing RLHF models on the Vicuna benchmark. To make this technology accessible to developers, NVIDIA has released STEER-LM as open-source software within its NVIDIA NEMO framework. Developers can now access the code and try STEER-LM with a customized 13B Llama II model. Overall, this development is a significant step towards creating more customized and adaptable AI systems.

OpenAI Considers Graphcore Acquisition

Why OpenAI Should Acquire Graphcore

Companies are focused on developing advanced algorithms and creating their own AI chips in AI. OpenAI, led by Sam Altman, is one such company planning to venture into the hardware market. However, they are facing a challenge—securing a reliable supply of AI chips. To address this issue, OpenAI is considering acquiring or investing in GraphCore.

This startup has seen a decline in revenue and increased losses. GraphCore signed a deal with Microsoft in 2019 but faced setbacks when Microsoft decided to use NVIDIA chips instead of GraphCore’s for its cloud computing system. By acquiring or investing in GraphCore, OpenAI would secure a stable supply of AI chips and potentially form a transformative partnership.

This move could also depart from their current partnerships with Microsoft and NVIDIA. OpenAI isn’t alone in developing its own AI chips. Companies like Microsoft, Google, Amazon, and Apple also work on their own chip designs. AMD, Intel, and Cerebras are among other companies pursuing this path. While there is no guarantee that OpenAI will proceed with acquiring GraphCore as they assess other potential targets for investment or acquisition, it highlights the increasing importance of having control over the hardware aspect of AI development.

AI Chatbot Appointed as Headteacher

TOP £32,000-A-YEAR PRIVATE SCHOOL APPOINTS AN AI CHATBOT AS ITS NEW ‘HEADTEACHER’ IN UK FIRST

In a groundbreaking move, a top private school in the UK has appointed an AI chatbot as its new head teacher. Cotsmore School in West Sussex hired an artificial intelligence robot named Abigail Bailey to support the human headmaster. Abigail Bailey developed with machine learning and educational management knowledge, will advise on various issues, including supporting staff members and helping students with ADHD.

The AI technology is similar to ChatGPT, where users can type questions and receive answers from the chatbot’s algorithms. Cotsmore School has taken it further by providing individual AI robots to students to help them understand their learning styles. This move aims to teach students how to collaborate with artificial intelligence rather than replacing human teachers.

The headmaster believes that having the AI robot as a resource is calming and reassuring as it can provide immediate assistance without waiting for human input. The school’s forward-thinking approach extends beyond just hiring Abigail Bailey. They have also appointed another robot named Jamie Rayner for AI strategy and planning.

With these advancements, Cotsmore School is leading the way in embracing technology within education while still prioritizing the importance of human interaction.

Google’s Project Green Light Revolutionizes City Traffic

Google Says AI-powered System Is Improving City Traffic Flows

Google is revolutionizing city traffic flows with its AI-powered system, Project Greenlight. This innovative technology is used in 12 cities worldwide to optimize traffic flow and reduce emissions.

Using individual driver data from Google Maps, Project Greenlight creates models of intersections and analyzes traffic light interactions. The system then uses these models to adjust and improve traffic plans, ultimately reducing stop and start times at intersections.

Early results are promising, with stops at intersections reduced by up to 30% and emissions reduced by up to 10%. City engineers can easily monitor progress and see the full results of the changes within weeks.

One city that has already seen the benefits of this system is Kolkata, India. Officials there have praised Project Greenlight for improving traffic flow and reducing gridlock.

Google plans to expand Project Greenlight to more cities worldwide and invites city planners and engineers to join the program. With this AI-powered solution, Google aims to make our cities more efficient, less congested, and environmentally friendly.

Thank you for reading Hidden AI Daily!

That’s all for today’s issue of Hidden AI Daily! Remember, this latest AI and machine learning news newsletter is released daily. Your feedback is important to us, so please provide your thoughts on the mobile app. With your input, we can make this newsletter even better. Stay tuned for tomorrow’s edition!