All articles tagged with #agi
DeepMind CEO Demis Hassabis told Axios that AGI could arrive within the 2029–2030 window, framing the current era as a 'practice run' for a future of agentic AI and urging governments and industry to accelerate safety testing and governance as researchers pursue recursive self-improvement and more capable AI agents.
During Google I/O 2026, Demis Hassabis described the moment as the foothills of the singularity and unveiled Gemini for Science, a suite of tools in Google Labs and Antigravity aimed at accelerating scientific research and drug discovery, as part of Google's push toward AGI; Hassabis has previously said there’s a 50% chance of AGI by 2030, though he cautioned that it has not arrived yet.
Facing fears that a new model like Mythos could push AI toward AGI, the U.S. and China are reviving talks on an emergency communications channel and governance for AI; Washington seeks formal talks and safeguards, while Beijing views Western restrictions as traps and emphasizes diffusion of AI into practical applications, making a swift, comprehensive deal unlikely and signaling a longer, nuanced diplomacy ahead.
Microsoft and OpenAI have dropped the contract’s AGI clause, ending the exclusive, perpetual revenue-sharing tied to AGI. OpenAI can now sell its products on any cloud provider, a move that could help it court more enterprise customers ahead of a possible public listing, while Microsoft’s revenue from outside agreements continues. The revenue-share will now run only through 2030 with a cap, signaling a broader, less exclusive partnership and OpenAI’s push toward multi-cloud growth.
OpenAI unveiled a major update to its guiding principles, shifting away from a heavy emphasis on artificial general intelligence (AGI) in 2018 toward a broader focus on AI capabilities and iterative deployment. The new guidance signals a more competitive stance with rival labs, moving away from a stance of collaboration and restraint. It also softens commitments, offering more suggestions to the tech ecosystem rather than strict, company-bound promises, while framing AI governance as a democratic and societal concern that may require new economic structures and substantial AI infrastructure.
Fortune notes Nvidia CEO Jensen Huang’s claim that AGI has been achieved, a statement that collides with a growing push to define and quantify general intelligence. Recent work from Google DeepMind and Hendrycks–Bengio proposes a scientific framework (a 10-facet cognitive taxonomy) to evaluate AI across domains and compare to well-educated adults, highlighting a current “jagged” profile where models excel in some areas but lag in others. Other benchmarks like ARC-AGI and debates dating back to Turing illustrate how hard it is to define intelligence, while tech giants push AGI branding for marketing and financial aims (OpenAI/Microsoft contracts referencing profit thresholds) even as leaders like Altman caution that AGI is a sloppy term. Overall, experts agree there is no universal consensus on what AGI means or how to measure it, despite ongoing progress and hype in the field.
The piece argues that the apparent arrival of AGI will boost productivity and sustain ongoing GPU demand, making Nvidia the primary beneficiary of AI-driven growth. It initiates coverage with a Buy rating and a $208 price target, suggesting a favorable ~21x forward P/E despite near-term Iran-related risks that are viewed as temporary.
Jensen Huang sparked a debate by saying 'we've achieved AGI' in one interview, then tempered the claim by acknowledging current AI systems still require significant human guidance; at the same time he pushes a token-based approach to scale AI work, suggesting engineers should spend more on tokens and even proposing token-based compensation, a theme that surfaced in two interviews days apart as the industry weighs what qualifies as AGI.
Nvidia CEO Jensen Huang told Lex Fridman that AGI has been achieved, sparking excitement, but he later walked back the claim, cautioning that many AI agents bloom briefly before fading and noting OpenClaw’s viral presence and potential social applications built around AI agents.
NVIDIA CEO Jensen Huang claims AGI has arrived, but his reasoning rests on a narrow, test‑like definition that counts a single viral, monetizable AI moment as proof. While Fridman pushes for a broader, more transformative standard, Huang describes a scenario where an AI creates a viral app and monetizes briefly, then fades—hardly the kind of sustained, institutional intelligence people expect from AGI. The piece argues Huang’s conclusion shows how definitional flexibility can make a “yes, we’re there” answer easy, even if the real, long‑term impact remains uncertain.
A new benchmark called Humanity’s Last Exam aims to measure how close today’s AI models come to human-level knowledge by presenting 2,500 carefully vetted, PhD-level questions across 100+ subjects. Launched in 2025, it has been attempted by top models like GPT-4o, Google Gemini The top score reported so far is 48.4% (Gemini 3 Deep Think), far below typical human expert performance (~90%). The test prioritizes precise, non-searchable knowledge and verifiable answers, filtering out questions AI could answer via web search. While a high score would indicate expert-level capability in specific domains, researchers say it does not by itself signal AGI or autonomous, general intelligence.
Amazon plans to invest up to $50 billion in OpenAI, with $15 billion upfront and up to $35 billion contingent on OpenAI reaching artificial general intelligence or pursuing an IPO; the round could exceed $100 billion in total funding and values OpenAI at roughly $730 billion pre-financing.
OpenAI CEO Sam Altman warned in New Delhi that artificial general intelligence could arrive sooner than many expect and that the world is not prepared. He says progress is accelerating, with OpenAI aiming to build an intern-level AI research tool by September 2026 and a fully automated AI researcher by March 2028. A quicker pace could boost demand for data centers, chips, and cloud tools from companies like Nvidia, Microsoft, and Alphabet, while also flagging potential job losses and calling for global oversight to prevent over-centralization of AI technology.
In a Nature correspondence, Quattrociocchi, Capraro, and Marcus argue that Chen et al.’s claim that success in behavioural tests (including Turing-test variants) demonstrates artificial general intelligence is problematic. They present three grounds for skepticism, stressing that such performance reflects statistical pattern matching or task-specific competence rather than true general intelligence or understanding, and warn against equating behavioural mimicry with AGI.
AI safety expert Dr. Roman Yampolskiy warns that as early as 2027 up to 99% of human jobs could disappear due to artificial general intelligence and automation, highlighting a potential tectonic shift in economies, education, and policy. While many tasks may be automated, only a small set of human-centric roles may persist, sparking ongoing debate about which occupations survive and how society should adapt—potentially accelerating changes in training, safety nets, and employment strategies.