Google Expands AI-Powered Try-On Feature to Dresses

In partnership with

A Thought Leadership platform to help the world navigate towards Artificial General Intelligence We are committed to navigate the path towards Artificial General Intelligence (AGI) by building a community of innovators, thinkers, and AI enthusiasts.

Whether you're passionate about machine learning, neural networks, or the ethics surrounding GenAI, our platform offers cutting-edge insights, resources, and collaborations on everything AI.

What to expect from Towards AGI: Know Your Inference (KYI): Ensuring your AI-generated insights are accurate, ethical, and fit for purpose Open vs Closed AI: Get expert analysis to help you navigate the open-source vs closed-source AI GenAI Maturity Assessment: Evaluate and audit your AI capabilities Expert Insights & Articles: Stay informed with deep dives into the latest AI advancements But that’s not all!

We are training specialised AI Analyst Agents for CxOs to interact and seek insights and answers for their most pressing questions. No more waiting for a Gartner Analyst appointment for weeks, You’ll be just a prompt away from getting the insights you need to make critical business decisions. Watch out this space!!

Visit us at https://www.towardsagi.ai to be part of the future of AI. Let’s build the next wave of AI innovations together!

TheGen.AI News

Google Expands AI-Powered Try-On Feature to Dresses

Google has extended its 'Virtual Try-On' feature, initially launched for men’s and women’s tops, to now include dresses. This AI-powered feature allows users to see how items fit on various body types, taking into account details like shadows, wrinkles, and fabric draping. 

Users in the US can search for dresses and click on any style marked with a ‘try on’ badge. From there, they can view how the dress appears on a range of real models in sizes XXS to XXXL. Once they find a model that suits them, they can visualize the dress in more detail before being directed to the retailer’s website to make a purchase. The feature uses a generative AI technology designed specifically for virtual try-on (VTO), employing a diffusion-based technique to create high-quality, realistic images of garments on models.

However, the company faced challenges when applying this technology to dresses. Dresses are more complex garments, requiring careful attention to detail and covering more of the body, which complicates accurate virtual rendering.

This update is aimed at improving the online shopping experience by offering a more realistic preview of items before purchase. While virtual try-on tools are enhancing online shopping, many consumers will still prefer in-store shopping for certain items. Currently, this feature is available in the U.S., with potential plans for future expansion.

Accenture and Unilever Strengthen Partnership to Boost Agility with GenAI

Accenture has strengthened its strategic partnership with Unilever to streamline its digital infrastructure and integrate generative AI (GenAI) to boost business agility. The collaboration is part of a multi-year initiative aimed at setting a new industry benchmark for productivity through GenAI, focusing on scaling proven use cases that deliver cost savings and operational efficiency worldwide.

This partnership builds on previous efforts to explore how GenAI can be expanded across Unilever’s operations. Unilever has since concentrated on developing and testing new AI concepts and projects through its Horizon3 Labs. Hein Schumacher, CEO of Unilever, highlighted that the company has already implemented 500 AI applications, enhancing efficiency. As AI continues to evolve, Unilever sees further potential to improve performance and productivity with Accenture’s advanced tools, identifying areas where AI can have the greatest impact.

Additionally, Unilever will use Accenture’s GenWizard platform to speed up its technology and digital product development. Accenture's GenWizard platform, backed by over 350 patents, offers a comprehensive solution for achieving technology objectives. Julie Sweet, Accenture’s chair and CEO, emphasized that this next phase in their long-standing collaboration will enable Unilever to scale GenAI, driving productivity, uncovering new ways of working, and enhancing competitive advantage.

Pioneering AI-Driven Networks for the Future of Business Connectivity

Ruckus, known for its cutting-edge technology and customer-centric approach, remains optimistic about the potential of WiFi-7, a network solution designed to meet the increasingly complex demands of today’s business environment. With a growing decentralized workforce and unprecedented pressures from both businesses and the global economy, networks are critical to modern IT infrastructures. As technologies like AI, GenAI, and SD-WAN evolve rapidly, technology companies must keep pace by developing solutions that not only meet but enhance these advancements.

The networking industry, a key driver of this technological evolution, underpins seamless global operations and has seen significant growth. In India, the networking infrastructure industry was valued at $5.09 billion in 2023 and is expected to surpass $6 billion by 2028. Ruckus specializes in delivering purpose-built networks capable of thriving in even the most challenging environments.

WiFi-6, introduced in 2019, quickly became the go-to for high-performance industries due to its superior speed, network efficiency, and ability to handle crowded environments. Now, WiFi-7 promises even greater improvements, offering faster data rates, lower latency, and better interference management—making it ideal for advanced technologies such as AR/VR, 8K streaming, and the rapidly growing Internet of Things (IoT) ecosystem, according to Sanjiv Verma, Vice President – APAC, RUCKUS Networks. The WiFi-7 global market is projected to reach $24.2 billion by 2030, thanks to its ability to deliver exceptional user experiences, low latency, and support for advanced connected devices.

Today's networks must be scalable, AI-ready, flexible, intelligent, and secure, with WiFi-7 playing a key role in meeting the modern demands of CIOs, Sanjiv adds.

Ruckus's products are designed to address every organizational need, offering solutions across cloud and software management, switches, and access points. Jitendra Gupta, Regional Director for India & SAARC, notes that the enterprise-class WLAN market grew by 90% year-over-year, driven by sectors like education, hospitality, services, and retail. Cloud-managed wireless solutions are also gaining significant traction. The switching business in India saw 30.4% YoY growth, with industries like services, finance, telecom, and manufacturing contributing to this expansion.

In 2018, Ruckus launched the IoT Suite, an innovative approach to building networks that support IoT deployments. The company also introduced Cloudpath, a secure network access software allowing users and devices to securely join any network, wired or wireless. Sanjiv highlights the use of AI/ML-enabled analytics in their network controllers, which provide IT teams with insights to resolve network anomalies before they disrupt services.

Sanjiv also discussed the role of AI in modern networks, which have become more diverse and complex. With rising costs and the shift to hybrid work, Ruckus has developed new technologies to meet these challenges. AI is increasingly being used for network management, automation, troubleshooting, design, and optimization, aligning networks with business goals. Ruckus is capitalizing on the growing popularity of AI, particularly through its RUCKUS One platform, which features an AI engine for unified management, flexible subscriptions, and other benefits.

Ruckus prioritizes partnerships with channel partners, including distributors, system integrators, ISVs, and resellers, to meet the rising demand for networking solutions. Thomas Schwab, Worldwide Vice President of Channels and Alliances at Ruckus, emphasized the company’s commitment to fostering mutually beneficial relationships through the RUCKUS BIG DOGS Partner Program, which offers partners resources, training, and incentives based on their tier.

Jitendra Gupta reiterated that Ruckus is committed to delivering premium services across fast-paced industries such as hospitality, large conglomerates, and manufacturing, through its extensive suite of products, covering areas from edge computing to AIOps.

TheOpensource.AI News

HyperWrite Unveils Reflection 70B: The New Leader in Open-Source AI

Matt Shumer, co-founder and CEO of the AI writing startup HyperWrite, has introduced Reflection 70B, a new large language model (LLM) based on Meta’s open-source Llama 3.1-70B Instruct. This model utilizes a novel self-correction technique and has shown outstanding results in third-party benchmarks. Shumer revealed on the social platform X that Reflection 70B is now considered "the world’s top open-source AI model." 

The model has undergone extensive testing on benchmarks such as MMLU and HumanEval, with LMSys’s LLM Decontaminator ensuring the results are uncontaminated. Reflection 70B has consistently outperformed Meta's Llama models and competes with leading commercial models.

Although a demo of Reflection 70B is available on a "playground" website, heavy traffic has caused delays, and the team is working to secure more GPUs to accommodate the demand. Shumer emphasized that Reflection 70B not only competes with top-tier models but also introduces unique abilities, such as identifying and correcting its own errors.

As Shumer explained to VentureBeat, the idea for Reflection came from pondering how to teach LLMs to recognize and fix mistakes—a capability they typically lack. The model, named "Reflection," can assess its output for accuracy before delivering results to users.

Its strength lies in a technique called reflection tuning, which enables the model to detect and correct its reasoning errors during response generation. Reflection 70B includes special tokens for reasoning and error correction, allowing users to interact with the model in a more structured way. During inference, the model flags its reasoning in real-time, correcting mistakes before providing the final response.

The demo site suggests prompts like asking Reflection 70B how many "r" letters are in the word "Strawberry" or comparing the numbers 9.11 and 9.9—simple tasks that many AI models struggle with. While testing was slow, Reflection 70B eventually provided accurate answers after over 60 seconds.

All Hands AI Gains $5M in Funding to Create AI Agents for Developer Efficiency

Programming at its core is a creative process, but in today's environment of shifting responsibilities, much of a developer's time is consumed by what All Hands AI co-founder and CEO Robert Brennan refers to as “toil-oriented tasks,” such as writing unit tests, managing dependencies, and updating documentation. While AI may lack creativity, it excels at handling these routine tasks.

All Hands AI, which announced a $5 million seed funding round led by Menlo Ventures, aims to develop model-agnostic, open-source AI agents to take over these repetitive tasks, allowing developers to focus on more creative aspects of their work.

The goal of the OpenHands agent is to act as a proactive pair programmer, working closely with developers to handle many of the mundane tasks they face daily. This could include writing tests, deploying applications, and even recognizing when changes in one file, such as a function name, might affect other parts of the codebase, and suggesting adjustments accordingly.

“AI is set to revolutionize how developers work, but it won’t change their preference for open-source solutions, especially for tools that impact their daily tasks,” said Joff Redfern, partner at Menlo Ventures and former chief product officer at Atlassian. “By building in the open, All Hands is helping the software engineering community work toward an ideal AI-powered development experience.” 

Brennan, along with co-founders Xingyao Wang (chief AI officer) and Graham Neubig (chief scientist), brings extensive experience in natural language processing and building AI agents. Brennan previously worked on document summarization at Google and held executive roles in startups focused on machine learning and infrastructure. Neubig is an associate professor at Carnegie Mellon, specializing in natural language processing, and Wang has paused his doctorate at the University of Illinois Urbana-Champaign, where he researched interactive language agents powered by foundation models.

Brennan noted that while tools like Copilot are useful for developers, they aren’t yet designed to manage the entire “agentic loop of writing code” in the way a self-driving car functions. This is the long-term vision for All Hands AI, though it remains somewhat aspirational. It's not yet possible to simply assign an AI agent to a company’s entire JIRA backlog and have it complete every task. Brennan, like many in the industry, believes human developers will remain a crucial part of the process for the foreseeable future.

New Open-Source AI Model Corrects Its Own Mistakes to Prevent Hallucinations

When AI giants like Anthropic or OpenAI introduce upgraded models, they often capture widespread attention due to AI's growing influence on both personal and professional computing. AI seems to be advancing rapidly. However, a new AI model from New York-based startup HyperWrite is gaining attention for a different reason—it employs an innovative error-trapping system designed to reduce the common "hallucination" issues that frequently affect chatbots like ChatGPT or Google Gemini. Notably, earlier this year, Google Gemini mistakenly advised users to apply glue to pizza.

The new AI, named Reflection 70B, is based on Meta’s open-source Llama model, according to VentureBeat. HyperWrite plans to integrate this AI into its main product, a writing assistant designed to help users craft and adapt content to their needs—one of the creative tasks where generative AI excels.

What sets Reflection 70B apart is its error-detection and correction system, referred to as "reflection-tuning," which CEO and co-founder Matt Shumer calls the "world's top open-source AI model." As Shumer explained on X, while other generative AIs tend to hallucinate without realizing it, Reflection 70B is designed to spot its own mistakes and correct them before finalizing an answer. The system analyzes its outputs to identify errors and improve on them—hence the name "Reflection." It essentially runs its output back through the model, which is then tasked with identifying areas for improvement.

The concept of AI improving its own performance isn’t new. In April, Meta's Mark Zuckerberg suggested that the Llama model could train itself by trying different approaches, identifying correct outputs, and feeding them back into the model. Reflection 70B takes a more immediate approach to addressing AI hallucination and misinformation, providing real-time corrections rather than merely using corrected data for future training. To demonstrate this capability, Shumer shared an example where Reflection was asked how many "R"s are in the word "strawberry." While many AI models incorrectly answered "two," Reflection initially made the same mistake but then flagged the error and corrected itself, stating there are actually three "R"s.

AI accuracy and the prevention of misinformation are critical issues as more people rely on AI for news, opinions, and data. Ensuring that future AI systems align with humanity’s best interests is becoming increasingly important, as evidenced by recent agreements between the EU, U.S., and UK to ensure AI safety.

However, creating effective AI regulations is challenging, requiring a deep understanding of the complex math and logic that underpin AI models like ChatGPT and Reflection 70B. For example, new AI laws in California mandate disclosure if an AI is trained on computers capable of 10^26 floating point operations per second, or 100 septillion calculations per second. As lawmakers grapple with these technical challenges, we can only hope they handle the complexities of AI regulation better than ChatGPT handles counting "R"s in "strawberry."

TheClosedsource.AI News

OpenAI Co-Founder Ilya Sutskever’s New AI Startup SSI Secures $1 Billion for Safe Superintelligence

Safe Superintelligence (SSI), a new AI startup co-founded by Ilya Sutskever, the former chief scientist of OpenAI, has secured $1 billion in funding to develop advanced AI systems that surpass human intelligence, according to company executives in a statement to Reuters. SSI, which currently has a team of 10, will use the funding to acquire computing power and hire top talent, with research and engineering teams based in Palo Alto, California, and Tel Aviv, Israel.

While SSI did not disclose its valuation, sources close to the company revealed it to be worth $5 billion. This large investment highlights the confidence some investors still have in backing high-profile AI talent, despite the industry's recent downturn and the exodus of several startup founders to tech giants. Investors in SSI include prominent venture capital firms Andreessen Horowitz, Sequoia Capital, DST Global, and SV Angel, as well as NFDG, an investment group led by Nat Friedman and SSI’s CEO, Daniel Gross.

Gross emphasized the importance of having investors who support SSI's mission to focus on developing safe superintelligence. He noted that the company plans to dedicate several years to research and development before launching any products.

AI safety, which addresses concerns about preventing harm from rogue AI systems, has become a crucial topic amid fears that AI could act against humanity's interests or even lead to catastrophic outcomes. A proposed California bill seeking to regulate AI safety has divided the industry, with companies like OpenAI and Google opposing it, while others, such as Anthropic and Elon Musk’s xAI, are in favor.

Sutskever, a key figure in AI, co-founded SSI in June along with Gross, who previously led AI projects at Apple, and Daniel Levy, a former OpenAI researcher. Sutskever serves as chief scientist, Levy as principal scientist, and Gross handles computing resources and fundraising.

Sutskever's decision to launch SSI stems from his desire to explore new challenges after parting ways with OpenAI. In 2022, he was involved in OpenAI’s controversial decision to remove CEO Sam Altman, a move that was later reversed. Following the fallout, Sutskever left OpenAI in May, and his Superalignment team, which aimed to align AI with human values, was disbanded.

Unlike OpenAI’s unique corporate structure designed for AI safety, SSI operates as a traditional for-profit company. The startup is currently focused on building a team that aligns with its culture, with Gross noting that they prioritize candidates with strong character and extraordinary skills over traditional credentials.

SSI plans to partner with cloud and chip providers to support its computing power needs but has yet to finalize which companies it will collaborate with. Many AI startups work with firms like Microsoft and Nvidia for infrastructure support.

OpenAI Japan CEO Announces GPT-Next: 100x More Powerful Than GPT-4

In May, OpenAI introduced "GPT-Next" at an event in Paris, leaving many in awe. GPT-Next is seen as the successor to GPT-4, but there’s another story unfolding with "Project Strawberry" and a new model called "Orion" in development, which is linked to GPT-5. However, that's a topic for later discussion. Regarding GPT-Next, it promises enhanced multimodal capabilities and smarter training methods that could significantly boost AI functionality. At the KDDI Summit 2024, OpenAI Japan’s CEO, Tadao Nagasaki, announced that GPT-Next is coming soon and will be 100 times more powerful than GPT-4. Nagasaki shared that the model will use a scaled-down version of OpenAI's secretive "Project Strawberry," previously known as Project Q* or Q-star, which is part of OpenAI’s pursuit of Artificial General Intelligence (AGI). The improvements in GPT-Next are expected to stem from advancements in architecture rather than relying solely on more computing power. 

Nagasaki also revealed that, by the end of August, ChatGPT had surpassed 200 million active users, making it the fastest-growing software in history. He emphasized that no other platform has made AI so accessible in such a user-friendly way. Comparing GPT-3 and GPT-4, Nagasaki noted that AI is advancing exponentially, with OpenAI dedicated to speeding up AI integration into society.

OpenAI's progress in AI is accelerating rapidly. GPT-Next is expected to outperform GPT-4 by a factor of 100, with even greater advancements on the horizon with models like Orion and GPT-5 Strawberry. Back in May, OpenAI CTO Mira Murati introduced a model called GPT 4o, describing it as significantly faster than GPT-4. Murati stated, "GPT 4o represents the future of human-machine interaction," highlighting improvements in text, vision, and audio capabilities. GPT 4o is twice as fast, 50% cheaper, and has five times the rate limits of GPT-4 Turbo. Now imagine the leap with a model 100 times more powerful.

Murati also hinted at GPT-Next, suggesting it could surpass human intelligence in certain areas. Speaking at Dartmouth Engineering, she described the next-gen model as having PhD-level intelligence for specific tasks. According to Murati, GPT-3 had the cognitive abilities of a toddler, GPT-4 resembled a high school student, and the upcoming model will be even more advanced. She noted that users might find the new chatbot to be more intelligent than themselves in many interactions.

Unlock the future of problem solving with Generative AI!

If you're a professional looking to elevate your strategic insights, enhance decision-making, and redefine problem-solving with cutting-edge technologies, the Consulting in the age of Gen AI course is your gateway. Perfect for those ready to integrate Generative AI into your work and stay ahead of the curve.

In a world where AI is rapidly transforming industries, businesses need professionals and consultants who can navigate this evolving landscape. This learning experience arms you with the essential skills to leverage Generative AI for improving problem-solving, decision-making, or advising clients.

Join us and gain firsthand experience in how state-of-the-art technology can elevate your problem solving skills using GenAI to new heights. This isn’t just learning; it’s your competitive edge in an AI-driven world.

Transform the way you run your business using AI (Extended Labour day Sale)💰

Imagine a future where your business runs like a well-oiled machine, effortlessly growing and thriving while you focus on what truly matters.
This isn't a dream—it's the power of AI, and it's within your reach.

Join this AI Business Growth & Strategy Masterclass and discover how to revolutionize your approach to business.
In just 4 hours, you’ll gain the tools, insights, and strategies to not just survive, but dominate your market.

What You’ll Experience: 
🌟 Discover AI techniques that give you a competitive edge
💡 Learn how to pivot your business model for unstoppable growth
💼 Develop AI-driven strategies that turn challenges into opportunities
⏰ Free up your time and energy by automating the mundane, focusing on what you love

🗓️ Tomorrow | ⏱️ 10 AM EST

This is more than just a workshop—it's a turning point.
The first 100 to register get in for FREE. Don’t miss the chance to change your business trajectory forever.

Sign up here to save your seat! 👈

In our quest to explore the dynamic and rapidly evolving field of Artificial Intelligence, this newsletter is your go-to source for the latest developments, breakthroughs, and discussions on Generative AI. Each edition brings you the most compelling news and insights from the forefront of Generative AI (GenAI), featuring cutting-edge research, transformative technologies, and the pioneering work of industry leaders.

Highlights from GenAI, OpenAI, and ClosedAI: Dive into the latest projects and innovations from the leading organizations behind some of the most advanced AI models in open-source, closed-sourced AI.

Stay Informed and Engaged: Whether you're a researcher, developer, entrepreneur, or enthusiast, "Towards AGI" aims to keep you informed and inspired. From technical deep-dives to ethical debates, our newsletter addresses the multifaceted aspects of AI development and its implications on society and industry.

Join us on this exciting journey as we navigate the complex landscape of artificial intelligence, moving steadily towards the realization of AGI. Stay tuned for exclusive interviews, expert opinions, and much more!