Leading AI companies are rethinking their approach to large language models as scaling existing methods faces diminishing returns. OpenAI’s latest model, o1, represents a pivotal shift towards human-like problem-solving techniques.
The traditional focus on larger datasets and increased computing power is being reconsidered. Key figures, including former OpenAI co-founder Ilya Sutskever, highlight the plateauing benefits of scaling and call for more innovative methods. Power shortages, data scarcity, and high costs have also hindered the development of superior models like GPT-4.
New approaches like ‘test-time compute’ are gaining traction, enabling AI systems to evaluate multiple solutions before choosing the most suitable one. This advancement enhances model performance without requiring massive increases in computational resources. OpenAI, Google DeepMind, and others are rapidly adopting these techniques, marking a shift in the competitive AI landscape.
These advancements could significantly alter demand in the hardware market, challenging Nvidia’s dominance in AI chips. As AI evolves, companies are competing not only to improve models but also to redefine the tools and techniques shaping the future of artificial intelligence.
Tesla founder Elon Musk has revived a legal action against OpenAI, alleging the organisation abandoned its original non-profit mission. Filed in a California federal court, the amended complaint names Microsoft, Reid Hoffman, and Dee Templeton as defendants. Additional plaintiffs, including Shivon Zilis, a Neuralink executive and ally of Musk, have also joined the case.
Musk, a co-founder of OpenAI, accuses the organisation of exploiting Microsoft’s infrastructure in what his lawyers describe as a ‘de facto merger.’ He claims OpenAI has benefited from favourable treatment by Microsoft, disadvantaging competitors such as xAI, Musk’s AI venture. The lawsuit also raises concerns over alleged antitrust violations involving OpenAI board members and their connections to Microsoft.
The filing alleges Reid Hoffman and Dee Templeton facilitated agreements between OpenAI and Microsoft that violated antitrust laws. It further details how Hoffman’s dual roles at Microsoft and OpenAI may have allowed access to sensitive information. Zilis, a former OpenAI board member, expressed similar concerns internally but was reportedly ignored.
Musk’s lawyers argue that OpenAI’s transition to a profit-driven model undermines its foundational principles of transparency and safety. The complaint references incidents such as a 2018 cryptocurrency proposal that Musk vetoed, citing potential reputational harm. OpenAI has dismissed the lawsuit as baseless and characterised it as a publicity stunt.
AI companies, including OpenAI, are shifting away from the ‘bigger is better’ philosophy for training models. Instead, they are developing techniques that allow algorithms to ‘think’ in more human-like ways. These methods aim to address challenges such as massive energy consumption, hardware failures, and data scarcity that have hindered advancements in large language models.
OpenAI’s new model, o1, uses a technique called ‘test-time compute’, allowing it to consider multiple answers and choose the best option during use. This approach improves performance in complex tasks, like problem-solving and decision-making, without needing extensive pre-training. Noam Brown, an OpenAI researcher, revealed that even brief ‘thinking’ boosts the model’s capabilities significantly.
The industry-wide shift has broader implications for AI hardware, especially as Nvidia’s chips have been critical to AI training. Experts predict a move towards distributed cloud-based servers for inference tasks, potentially reshaping the demand landscape for chips. Prominent investors, such as Sequoia and Andreessen Horowitz, are monitoring these changes closely as they may impact investments in AI infrastructure.
Caitlin Kalinowski, previously Meta’s head of augmented reality (AR) glasses, has announced she will join OpenAI to lead its robotics and consumer hardware initiatives. Kalinowski, who managed Meta’s AR glasses and VR goggles divisions, is expected to leverage her expertise in hardware to advance OpenAI’s efforts in robotics and develop consumer-focused AI products. She will focus on bringing AI into the physical world through collaborative projects and new technology partnerships.
This move is part of OpenAI’s growing commitment to hardware. Recently, OpenAI teamed up with Jony Ive’s LoveFrom to design a consumer AI device aimed at creating a computing experience that minimises social disruption. OpenAI has also resumed hiring robotics engineers after a previous shift away from hardware, reflecting a renewed interest in integrating its AI models into physical applications.
Kalinowski joins at a time when several companies, including Apple, are beginning to integrate OpenAI’s AI models into consumer technology. With the addition of Kalinowski, OpenAI aims to bring advanced AI functionality into robotics and consumer devices, promising transformative new products.
Physical Intelligence, a robot AI startup, has raised $400 million in early funding, attracting high-profile investors like Jeff Bezos, OpenAI, Thrive Capital, and Lux Capital. This latest investment values the startup at $2 billion, underscoring strong interest in AI-driven robotics solutions. Physical Intelligence aims to develop foundational software adaptable across different types of robots, eliminating the need for custom software per specific robotic task.
Global technology giants like Microsoft, Google, Meta, Amazon, and Nvidia are already funnelling billions into AI. Industry estimates from venture capital firm Accel project that AI and cloud tech funding across the US, Europe, and Israel will hit $79.2 billion by the end of 2024. With several startups joining the robotic AI sector, including Vicarious, Universal Robots, and Covariant, competition in this space is intensifying.
Elon Musk recently predicted that at least 10 billion humanoid robots will be available by 2040, with prices ranging from $20,000 to $25,000. His Tesla company showcased the latest Optimus humanoid robot, adding to the anticipation around robotics advancements and automation technologies. Musk’s vision is part of the broader trend pushing the potential for AI-powered robots in everyday life.
Physical Intelligence recently demonstrated its innovative software, named π0, or pi-zero, showing its flexibility by enabling robots to carry out complex household tasks like folding laundry, bagging groceries, and retrieving toast from a toaster. This achievement highlights the startup’s strides toward building a universal robot software platform, with potential impacts on industries from home automation to logistics.
OpenAI, led by Sam Altman, is exploring a move to transition from a non-profit to a for-profit company, according to Bloomberg News. Early discussions with the California attorney general’s office aim to alter OpenAI’s corporate structure, signalling a significant shift in the governance of the AI research pioneer. Originally founded in 2015 as a non-profit, OpenAI’s new direction could open the door to increased investment opportunities.
The potential change in status would mark a strategic shift for the company, which developed the popular AI tool ChatGPT. OpenAI’s connection with Microsoft, which invested heavily in the company, has been a key factor in its growth. In September, reports emerged that OpenAI was already contemplating restructuring as a for-profit benefit corporation, with a new structure aimed at enhancing business opportunities while keeping its non-profit arm involved.
The non-profit branch of OpenAI would continue to exist and hold a minority stake, allowing the organisation to maintain some of its founding mission. This balance could appeal to investors while retaining a foothold in its original purpose of ethical AI development. OpenAI’s funding has also surged recently, including a $6.6 billion funding round, potentially valuing the firm at $157 billion.
With substantial backing and increased investor interest, OpenAI’s shift could cement its position as one of the world’s most valuable private companies. The move reflects broader trends in AI, where companies are aligning profit goals with technological innovation to remain competitive in a rapidly evolving market.
Apple is set to integrate OpenAI’s ChatGPT with Siri and other iOS features, rolling out in December with iOS 18.2. Beta testers of iOS 18.2 have discovered an option in the Settings app allowing users to subscribe directly to ChatGPT Plus, OpenAI’s premium plan, for $20 a month. This move offers Apple users a streamlined way to access ChatGPT‘s advanced features, which include more capable AI models and additional voice and image options.
Free users of ChatGPT will still face some limitations, including restricted access to OpenAI‘s most advanced models and fewer image generations per day. However, OpenAI could see a surge in paid subscriptions from the partnership as Apple’s exposure could drive substantial user interest in ChatGPT Plus.
The specifics of the financial terms between Apple and OpenAI are unclear. Apple isn’t reportedly paying OpenAI directly, relying instead on enhanced exposure, though questions remain about any potential revenue sharing. Apple’s broader AI plans indicate it will also include models from other developers in future updates, possibly incorporating Google’s Gemini.
In the background, OpenAI is working to secure additional funding amidst a period of rapid change, including the departure of key executives and a recent $6.6 billion funding round, which Apple reportedly declined to join following unexpected shifts at OpenAI.
OpenAI has introduced new search functions to its popular ChatGPT, making it a direct competitor with Google, Microsoft’s Bing, and other emerging AI-driven search tools. Instead of launching a separate search engine, OpenAI chose to integrate search capabilities directly into ChatGPT, which will pull information from the web and relevant sources based on user questions.
Initially, ChatGPT’s search feature will be available to Plus and Team users, with plans to expand access to enterprise and educational users, as well as free users, in the coming months. OpenAI’s partnerships with major publishers like Condé Nast, Time, and the Financial Times aim to provide a rich pool of content for ChatGPT’s search.
This launch follows OpenAI’s selective testing of SearchGPT, an AI-based search prototype, earlier in the year. With its recent funding round boosting its valuation to an estimated $157 billion, OpenAI continues to strengthen its standing as a leading private AI company.
OpenAI is reportedly set to launch a powerful new AI model, code-named Orion, with an initial release expected by December. Unlike its predecessors, Orion will be selectively available at first, with trusted partner companies given early access to integrate the model into their products. OpenAI’s primary partner, Microsoft, is preparing to host Orion on its Azure platform as early as November.
While some within OpenAI view Orion as a successor to GPT-4, it is unclear whether it will be formally named GPT-5. OpenAI has not confirmed the launch date, and CEO Sam Altman recently downplayed the existence of Orion. Nonetheless, speculation continues as an executive hinted that Orion may be up to 100 times more powerful than GPT-4, moving the company closer to its ambitious goal of artificial general intelligence.
Reports suggest that synthetic data from OpenAI’s 0.1 model, released earlier this year, helped train Orion. OpenAI has teased the model’s arrival through cryptic social media posts, with Altman recently referencing the upcoming “winter constellations” — a possible allusion to Orion, a prominent winter constellation.
Orion’s anticipated release aligns with OpenAI’s completion of a $6.6 billion funding round, with restructuring towards a for-profit model. The company, however, is facing notable internal changes, including the recent departures of CTO Mira Murati and other key research leaders, amid the heightened focus on this next-generation AI model.
Miles Brundage, a veteran policy researcher and senior adviser at OpenAI, has left the company to pursue independent work in the nonprofit sector. In a post on X and an essay, Brundage explained his decision, stating he believes he can have a greater impact on AI policy and research outside of the industry, where he will have more freedom to publish his findings.
Brundage joined OpenAI in 2018 and played a key role in the company’s policy research, particularly in the responsible deployment of AI systems like ChatGPT. His departure signals ongoing shifts within OpenAI, with the company reorganising its economic research and AGI readiness teams. While OpenAI expressed support for Brundage’s decision, it did not specify who will take over his responsibilities.
Brundage’s exit is part of a broader trend of high-profile departures from OpenAI, with several key figures, including CTO Mira Murati and chief research officer Bob McGrew, having recently resigned. The departures reflect internal disagreements about the company’s direction, especially as it faces criticism over balancing commercial ambitions with AI safety.
