Former Google AI executive Jad Tarifi says Integral AI has achieved human-level AGI, but researchers say definitions and evidence remain unclear.
The world’s richest man, Elon Musk, recently claimed his artificial intelligence company, xAI, could achieve artificial general intelligence by 2026. This is thanks to the firm’s Grok 5 model, which it plans to release early in the year.
AI enthusiasts may have felt a sense of deja vu. Roughly a year ago, Musk said xAI would achieve AGI in 2025.this year. Much like Musk’s assertions, it’s important to take these groundbreaking milestone claims with a pinch of salt. When a technological milestone isn’t widely understood, it’s easy for companies to claim responsibility for a historic achievement. So what is AGI? Much like Musk, Tarifi claims that his company’s version of AGI will serve a higher purpose, improving the human experience on a global scale. “We’re entering an era of universal freedom,” Tarifi toldWhat is AGI, though? Artificial general intelligence refers to a hypothetical type of AI that possesses human-level cognitive abilities. This means it would be able to understand, learn, and conduct intellectual tasks in the same way a human being can. Right now, AI models are designed for specific, limited tasks, making them incredibly efficient at a narrow set of skills. However, when it comes to learning and applying new skills to a broad set of tasks—the way humans do—they are much more limited. It’s hard to precisely describe what it means to achieve this level of human cognition. As such, Integral AI provided its own definition of AGI. In its statement, the company claims that AGI is determined by meeting three fundamental qualifiers: autonomous skill learning, safe and reliable mastery, and energy efficiency. Autonomous skill learning refers to a system that can “teach itself entirely new skills in novel domains without pre-existing datasets or human intervention,” the company explained. Safe and reliable mastery, meanwhile, describes learning without “catastrophic risks or unintended side effects.” Lastly, energy efficiency refers to the energy cost of learning being roughly equivalent to that of a human acquiring the same skill. Integral AI claims it used these principles as “fundamental cornerstones and developmental benchmarks during the inception and testing of this first-in-its-class AGI learning system.” In an interview with IE, Integral CEO Jad Tarifi elaborated further: “We define AGI pragmatically: a system that can learn any skill autonomously, reliably, safely, and efficiently. This is a practical definition because it means any task can eventually be solved more cheaply than with human involvement.” “What we’ve built is a world model with fundamentally better abstractions,” he continued. “It learns continuously, plans efficiently, generalizes well, and takes safe actions to collect its own training data. We call this paradigm “interactive learning.” We’ve demonstrated it across five environments and are now scaling further.”Of course, it’s hard to quantify or substantiate such an achievement. It’s also hard not to raise an eyebrow when a company provides its own definition for a milestone it claimed it has achieved.Still, Tarifi claims AGI will be “utterly transformative,” and his company is currently at the forefront of this technological leap. According to Tarifi, who was the founder of Google’s first generative AI team, the narrative he cares about most is this: “We’re entering an era of universal freedom. Human beings will increasingly be able to author their lives in whatever way they choose. The question is whether we design our systems so that freedom is broadly distributed.”on Tarifi and Integral AI. The article claimed that the Google veteran was discouraging people from pursuing degrees in law or medicine. By the time any student finishes one of these years-long degrees, their field will have completely changed due to AI. “That’s an interesting example of how ideas get recontextualized in the media,” Tarifi said when asked about the article. “My actual point was more nuanced: education will evolve, as it always has and always will. I generally discourage people from pursuing PhDs unless they’re genuinely passionate about a topic and willing to accept significant personal sacrifice. I similarly discourage people from starting companies.” “A PhD in AI specifically may be particularly challenging given the field’s pace; if someone wants to contribute meaningfully, working in industry often provides faster feedback loops and more immediate impact,” Tarifi continued. “That’s not the same as saying education is obsolete. It’s saying the optimal path to learning and contribution is developing.”Tarifi claims that AI and AGI will allow humans to focus on the things that are truly important: fostering important relationships and prioritizing human connection. Paradoxically, the technology, typically associated with robots, could free humans to simply be human. “Understanding human connection isn’t just something we’ll have more time for,” Tarifi told IE. “It’s fundamentally empowering. It’s the kind of capability that’s both intrinsically valuable and will become even more valuable post-AGI. As routine cognitive tasks get automated, our capacity for deep relationship and meaning-making with others becomes even more foundational to a flourishing life.” Of course, the other side of the argument is that AI could put immeasurable power in the hands of the few. OpenAI was famously founded as a non-profit to provide AI research and capabilities to the masses. However, the organization announced the completion of its conversion to a fully for-profit public benefit corporation in October this year.Tarifi understands that his company’s assertions will be met with scepticism. In fact, the Integral CEO claims he understands where the scepticism comes from. “Current large language models have fundamental limitations. They don’t understand things reliably, even when they appear eloquent. The fluency can mask a lack of genuine comprehension,” he said. “Where I differ from sceptics is in believing these limitations aren’t inherent to AI itself, but to the current paradigm. Our approach at Integral doesn’t suffer from those same limitations because we’re building systems that learn through interaction with the world.” According to renowned University of Oxford AI expert Professor Michael Wooldridge, seeing is believing. “One has to be very careful about press releases these days in AI because the level of hype is so high,” Wooldridge told IE in an interview. “For any grand announcement, we would really need to see fully authenticated details.” In the specific case of Integral AI, Wooldridge said his “gut reaction is to be rather sceptical.” This is because “there is no clear-cut consensus on exactly what AGI means. For some people, it means having AI that could learn to do anything that a human being could learn to do, which is an extraordinarily ambitious goal.” “I am very, very sceptical that we have AI at that level,” Wooldridge continued. “We don’t even have AI that could safely drive me from Oxford to London, nor do we have AI that could come into my house and cook a meal to order. How could we possibly have AGI if we don’t have AI that can do those things?” Like Musk, Tarifi is comfortable making sweeping statements. In Integral’s press release, he said the company’s work “marks the next chapter in the story of human civilization.”Time will tell whether Integral will truly roll out a revolutionary AGI platform in the near future to substantiate its own recent claims. While boosting human connection is an ambitious and noble goal, we have yet to see any substantiated evidence that generative AI provides an efficient means to achieve this goal.Chris Young is a journalist, copywriter, blogger and tech geek at heart who’s reported on the likes of the Mobile World Congress, written for Lifehack, The Culture Trip, Flydoscope and some of the world’s biggest tech companies, including NEC and Thales, about robots, satellites and other world-changing innovations.AI and RoboticsAI and Robotics
United States Latest News, United States Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
NYPD brass retaliated against former captain for reporting corrupt promotions, lawsuit claimsA former New York Police Department captain is suing the department and outgoing Mayor Eric Adams, accusing former police executives of placing “cronies” who
Read more »
Former Congressman Rob Bishop poised to return to the Utah House, where his political career beganFormer Congressman Rob Bishop is looking to return to the Utah House, where his political career began more than 50 years ago.
Read more »
Arraignment delayed again for former CBI scientist accused of mishandling DNA evidenceRobert Garrison is a Colorado native who grew up in Montrose and attended Colorado Mesa University. He is an AP award-winning journalist who joined Denver7 in August 2016 after working in CO and OK.
Read more »
Google Workspace Promo Code: Up to 14% Off in 2026Save on your Google Workspace plan with verified promo codes from WIRED. Get 10% off Starter, 12% off Standard, or 14% off Plus plans for three months when you sign up before January 31, 2026.
Read more »
Google’s parent buys data center energy specialist Intersect for $4.75 billion to help powThe news comes a month after the search giant unveiled plans to invest $40 billion in Texas, including three new data centers across the state.
Read more »
How Google, Microsoft, Walmart, are preparing for an aging workforceBusiness Insider tells the global tech, finance, stock market, media, economy, lifestyle, real estate, AI and innovative stories you want to know.
Read more »