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. Each edition brings you the most compelling news and insights from the forefront of General 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 OpenAI, the organization behind some of the most advanced AI models like GPT and DALL-E. Equally, we'll give you an inside look at the initiatives and strategies from ClosedAI, a hypothetical entity committed to pushing the boundaries of closed-source AI development.

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!

TheGen.AI News

SAP and IBM Consulting Forge Partnership to Launch Generative AI Services

IBM’s consulting branch is teaming up with SAP to deliver generative AI services aimed at speeding up digital transformation for enterprises through the RISE with SAP program, according to a recent announcement.

The collaboration, revealed on Wednesday, centers around deploying generative AI services through RISE with SAP. This move builds on a previous May partnership where SAP incorporated IBM's Watson AI into its software suite, including S/4 HANA and SAP Business One, enhancing these solutions with Watson's natural language processing and predictive analytics.

The latest initiative aims to develop and offer AI-driven applications and services for enterprise business processes and platform architecture, leveraging the capabilities of SAP's cloud solutions under the SAP Business Technology Platform (SAP BTP).

IBM will focus on integrating AI enhancements across SAP’s cloud-based applications, touching on multiple domains like finance, supply chain, human resources, customer experience, and spend management. They also plan to utilize SAP Signavio and Business AI to innovate business process design through a series of proof-of-concept programs.

Furthermore, IBM has started crafting an extensive range of over 100 AI-driven solutions spanning various industries and business functions, which will be accessible globally via IBM Innovation Studios and SAP Experience Centers.

To facilitate enterprise adoption, IBM and SAP have structured a Value Generation initiative, where IBM will develop next-generation reference architectures using SAP BTP and Signavio, aiming to establish standards in data management, system processes, and automation.

Virtusa Debuts Helio, Enhancing Enterprise Outcomes with Generative AI

Virtusa Corporation, a leader in digital strategy, engineering, and IT services, has unveiled Virtusa Helio, a comprehensive set of platforms, accelerators, and consulting services designed to help organizations effectively utilize generative AI (genAI) in their digital transformation efforts. Emphasizing Virtusa’s “Engineering First” philosophy, Helio aims to guide enterprises through their genAI adoption processes, focusing on scalable business solutions.

This introduction responds to the growing demand for resources that aid enterprises in understanding and integrating genAI technologies amidst a fast-evolving landscape.

According to Gartner, by 2030, AI is expected to influence every dollar of global GDP, daily AI interactions for every individual in developed countries, and the incorporation of AI in all IT development.

Virtusa's research indicates a widespread optimism towards genAI, with 90% of survey participants believing AI will enhance their organization’s efficiency. However, despite the eagerness for adoption, many find the actual implementation challenging.

Santosh Thomas, CEO of Virtusa, noted the rapid uptake of genAI, comparing its revolutionary impact to the internet's emergence. He stated that while many enterprises are keen to leverage genAI, integrating it effectively with their ongoing digital strategies remains a significant hurdle. The Helio suite is engineered to help organizations fundamentally transform and capitalize on this new technological era.

The Virtusa Helio suite includes:

- Helio Innovate: A platform for genAI and AI experimentation and prototyping.

- Helio Build: Tools aimed at boosting productivity and efficiency in application development.

- Helio Enhance: Solutions for accelerating the modernization of legacy systems using genAI.

- Helio Operate: Solutions to enhance the management and operation of software systems and infrastructure with genAI.

Additionally, Virtusa has established a generative AI Center of Excellence with over 450 engineers specialized in various aspects of genAI. This center offers:

- Consulting and advisory services including ideation, strategy development, and governance.

- Development of tailored solutions, from experimentation to productization and data services.

- Assurance services to certify genAI implementations and ensure their effective operation.

- AI-assisted engineering to improve developer productivity and support legacy system modernization.

Virtusa Helio integrates these advanced software tools with Virtusa’s deep engineering expertise to provide a robust resource for companies aiming to swiftly yet methodically embrace genAI, ensuring long-term business success.

TCS Ramps Up Generative AI Capabilities, Projects Stronger FY25 Economic Outlook

Tata Consultancy Services (TCS) is advancing its position in Artificial Intelligence (AI), particularly in generative AI, as highlighted by CEO K Krithivasan and chairman N Chandrasekaran in their communications to shareholders. They also shared a positive economic forecast for the 2025 fiscal year.

Krithivasan discussed TCS's leadership in applying generative AI to boost productivity and make significant strides across various industries. In FY24, TCS merged its AI and cloud departments into a single AI.Cloud unit, which facilitated the training of 300,000 employees in genAI technologies. The company has integrated AI functionalities into its software offerings, such as TCS Bancs for banking and various cybersecurity products, enhancing both efficiency and security.

In the financial sector, AI and genAI have notably improved the speed of insurance claims processing, lending, and disbursements. These technologies have also allowed TCS's primarily financial clients to better utilize data for personalizing products and services, thus enhancing customer engagement and acquisition.

Additionally, Krithivasan emphasized TCS's contributions to other sectors. The company is aiding the energy transition with AI-driven solutions for green hydrogen and biofuels and developing digital platforms to help reduce greenhouse gas emissions. In healthcare, AI is speeding up drug development and transforming care delivery by increasing productivity, enabling remote care, and addressing the shortage of skilled professionals. In the manufacturing realm, AI tools like machine learning, computer vision, and augmented reality are cutting costs and boosting safety and efficiency.

TCS is also at the forefront of the telecommunications industry's migration to 5G and 6G, developing communication stacks based on open standards and conducting R&D in advanced telecom technologies in Bengaluru and Gurugram. The focus includes proprietary technologies in 5G SA Core and Open Radio Access Networks (O-RAN) radios.

On the business front, Krithivasan noted that clients are focusing on projects with quick returns and are demanding that transformations be funded through operational savings. The main business drivers are cost optimization and cloud transformation, propelled by vendor consolidation, cloud migration, and enhancements in customer and employee experiences.

Chandrasekaran shared an optimistic macroeconomic perspective, noting positive trends such as growth, disinflation, and potential monetary easing, despite ongoing geopolitical challenges affecting global supply chains. He pointed out that these challenges are prompting companies to reconfigure their supply chains for increased resilience and efficiency, with India emerging as a key player in the new global supply chain ecosystems.

TheOpen.AI News

Open Source at the Forefront of AI Innovation, Says Red Hat CEO

Matt Hicks, the CEO of Red Hat, recently highlighted the significant role of the open source ecosystem in advancing artificial intelligence (AI) during his keynote speech at the Red Hat Summit. According to Hicks, AI development will be characterized by a variety of options, rather than being dominated by any single vendor. He noted that innovation is rapidly progressing, with new advancements emerging constantly.

Hicks emphasized the critical intersection of open source and academic research in fueling AI development. He pointed out that some of the most significant AI capabilities are being developed by top academic researchers, and open source is essential for unlocking this potential.

During the last year, Hicks observed a surge in the diversity of AI models, stating that capabilities once limited to very large models can now be achieved with significantly smaller ones. Moreover, these capabilities, which previously required advanced hardware, can now operate on standard laptops. Additionally, AI training costs have dramatically decreased.

Red Hat, acquired by IBM in a $34 billion deal in 2018, is placing strong emphasis on AI. This year's Red Hat Summit is focused on the combination of open source and AI, which Hicks believes will continue to have a profound impact as AI technology becomes more accessible and cost-effective.

Hicks foresees a future where AI is developed collaboratively across global institutions, from MIT to IITs in India, fostering a cycle of continuous innovation. He predicts that open source will be the foundation for AI deployment across various platforms. The summit also included announcements of new partnerships and collaborations by Red Hat.

IBM Launches InstructLab, Opens Up Granite AI Models to Public

IBM has launched a significant initiative into the open-source development of large language models (LLMs) with its new AI project, InstructLab, developed in collaboration with its software subsidiary, Red Hat. Announced on Tuesday, InstructLab aims to offer a model-neutral environment where anyone can contribute to and learn from the ongoing development of LLM infrastructures. This launch aligns with IBM's release of its Granite models into the open source, which were previously exclusive to commercial applications.

Generative AI applications heavily depend on LLMs, but most available models are proprietary, such as OpenAI's GPT models, or less open than claimed, like Meta's Llama models. This often forces developers to replicate and modify an existing LLM's architecture and codebase, a process called "forking," which requires significant investment without the ability to contribute improvements back to the original model.

This exclusivity also presents broader economic implications, as the dominance of a few major tech firms over essential LLMs could restrict innovation. In response to these concerns, Mozilla and nearly 50 other organizations and academics called on the U.S. Department of Commerce in March to support open-source AI models to prevent market domination by these few entities.

IBM's InstructLab is designed to counter this closed ecosystem by providing an open platform for LLM development. The initiative builds on IBM's Granite models, a suite of foundation models under its IBM Watsonx brand, launched in September 2023. These models, including Granite.13b.instruct and Granite.13b.chat, utilize a decoder architecture ideal for applications in various fields such as academia, law, finance, development, and the web, but were initially available only to Watsonx subscribers.

Now, IBM is democratizing access to its enhanced Granite 7B model, a reference implementation of Meta's Llama-2-7B, and has open-sourced additional models like Labradorite and Merlinite. IBM's aim with InstructLab is to make it easier for developers to engage with LLMs. For enterprises still needing specialized solutions, Red Hat has introduced Enterprise Linux, a new open hybrid cloud platform for large-scale business applications.

Startup Tenyx Takes the Lead with Fine-Tuned Llama 3 Outperforming GPT-4

In a recent interview with VentureBeat, Itamar Arel, CEO of AI startup Tenyx, announced a significant advancement in natural language processing. His company has enhanced Meta’s open-source Llama-3 language model, now rebranded as Tenyx-70B, enabling it to surpass OpenAI’s GPT-4 in specific areas. This marks a historic achievement for an open-source model over a proprietary one.

Arel described Tenyx's fine-tuning method as a way to refine a foundational model further than its original training, allowing for what he termed "continual learning" or "incremental learning." This approach addresses the problem of "catastrophic forgetting," where a model forgets its prior knowledge upon learning new information. By modifying only about 5% of the model's parameters, Tenyx manages to update the model without losing its existing capabilities and achieves remarkably rapid training times—fine-tuning the 70-billion-parameter model in just 15 hours with 100 GPUs.

Tenyx's decision to release their optimized model under the same open-source license as the original Llama-3 underscores their commitment to open-source AI, which Arel believes benefits the entire community by facilitating more applications and innovations.

The applications of Tenyx's technology are extensive, from developing industry-specific chatbots to allowing more regular updates to AI models, thus keeping them up-to-date between major upgrades. This breakthrough could democratize access to cutting-edge language models, previously dominated by proprietary models, potentially spurring further advancements within the open-source community.

Arel also reflected on the broader implications of their success, questioning its impact on leading AI firms like OpenAI. This development could significantly influence the AI landscape, shifting how businesses engage with natural language processing tasks.

Although the optimized Llama-3 model, Tenyx-70B, maintains some of the base model's limitations, such as producing occasional illogical responses, it has shown remarkable improvements in areas like mathematics and reasoning, with accuracy improvements from 85% to nearly 96%.

Tenyx's breakthrough opens up new possibilities for open-source AI, challenging the dominance of proprietary models and setting the stage for a future where open collaboration and innovation lead the way in artificial intelligence.

TheClosed.AI News

OpenAI's CEO Sam Altman Discusses AI's Threat to Employment

OpenAI CEO Sam Altman recently expressed his concerns about the rapid socioeconomic changes driven by Artificial Intelligence (AI) and its potential impacts, particularly on elections and the job market. Speaking at a Brookings Institute panel on AI and geopolitics, Altman highlighted the dangers of AI-generated misinformation in elections and noted that discussions about AI’s economic effects have dwindled since last year, indicating that these concerns are not being addressed with the seriousness they warrant.

He pointed out that while the release of GPT-4 did not show a significant detectable impact on the economy, this should not lead to complacency about the potential issues AI could cause in the future. "There's a risk that we won't take it seriously enough moving forward, and it’s a massive issue," he remarked.

This discussion follows a study by the International Monetary Fund (IMF) suggesting that AI could automate nearly half of the jobs in advanced economies, which could result in reduced hiring and lower wages.

In a prior interview with CNBC, Altman admitted to having some reservations about the capabilities of ChatGPT, OpenAI's popular AI model, including its potential to "eliminate" many jobs. "If I said I were not [a little bit scared], you should either not trust me, or be very unhappy that I'm in this job," he had stated.

OpenAI Launches Long-Awaited Deepfake Detection Tool

OpenAI recently launched a deepfake detector specifically designed to identify images created by its Dall-E 3 generative AI system, receiving a broadly positive response despite its proximity to the U.S. presidential election just five months away. This tool arrives amid concerns about the potential use of deepfakes to manipulate voter behavior in upcoming global elections.

Kashyap Kompella, CEO of RPA2AI Research, welcomed the introduction, highlighting the complex challenge of detecting AI-generated content and noting that it requires efforts beyond a single company.

During the 2020 U.S. elections, deepfakes were notably used for political misinformation. Since then, the sophistication and accessibility of generative AI technologies have significantly advanced, increasing the potential for misuse. Given this backdrop, OpenAI and other leading GenAI companies like Stability.ai, Midjourney, and Adobe have faced growing scrutiny to ensure their systems are secure and trustworthy.

The new OpenAI deepfake detector claims a 98% accuracy rate and is currently available to selected testers. However, it remains unclear how broadly it will be used and whether other GenAI firms will develop similar tools.

Beyond the detector, OpenAI has also taken additional steps to combat misinformation. It has established a $2 million education fund in partnership with Microsoft to promote AI literacy among older adults and other groups vulnerable to deepfake deception. Moreover, OpenAI has joined the steering committee of the Coalition for Content Provenance and Authenticity (C2PA), an organization setting technical standards to help identify the origins of media content and combat misinformation.

Critics like Alon Yamin, CEO of Copyleaks, acknowledge the detector as a positive move but caution that it addresses only content from OpenAI’s platforms, leaving a gap in detection capabilities across other generative AI systems capable of creating deceptive content.

Despite these concerns, some experts, including Forrester Research analyst Audrey Chee-Read, view the timing of the release as strategic, coinciding with the critical final stages of election campaigning. This period is crucial for influencing voter decisions, making it a pertinent time for such a tool to be introduced.

Microsoft Steps Up in AI Race with Mustafa Suleyman and New MAI-1 Model

Microsoft is advancing its position in the AI landscape with the development of a new AI model named MAI-1, led by former Google AI chief Mustafa Suleyman. This new model, as reported by The Information, is part of Microsoft's ambitious efforts to compete with major AI players like Google, Anthropic, and OpenAI. MAI-1 is touted to have around 500 billion parameters, which is less than OpenAI's GPT-4 but still represents a substantial leap in terms of learning capabilities during the AI training phase.

The creation of MAI-1 follows Microsoft's acquisition of the majority of the team and intellectual property from Inflection, an AI startup co-founded by Suleyman, for USD 650 million. While MAI-1 is a separate initiative, it is likely to benefit from the technologies and training data acquired from Inflection.

Mustafa Suleyman, a prominent figure in the AI world, was one of the co-founders of DeepMind, which was acquired by Google in 2014. He has been influential in shaping AI policy and product development at Google before leaving to start Inflection AI with Reid Hoffman in 2022. His move to Microsoft as AI CEO earlier this year was celebrated by Microsoft CEO Satya Nadella, highlighting the significance of his role in driving Microsoft's AI strategies.

Suleyman's transition to Microsoft and the development of MAI-1 mark significant milestones in Microsoft's AI endeavors, underscoring its commitment to maintaining a competitive edge in the rapidly evolving AI industry.

Join the movement - If you wish to contribute thought leadership, then please fill the form below. One of our team members will be back in touch with you shortly.

Form link

Keep reading