- Microsoft AI CEO says AGI will be achieved within 5-7 years.
- Unlike OpenAI’s CEO, the benchmark requires sophisticated hardware.
- Microsoft’s OpenAI partnership could be a historic success.
Several reports have emerged in recent months hinting at the possibility that AGI (artificial general intelligence) will be achieved sooner than expected. On the one hand, Anthropic CEO Dario Amodei speculates that the benchmark could be reached between 2026 and 2027, while OpenAI CEO Sam Altman narrows it down. the calendar until 2025.
Contrary to popular belief, Altman states benchmark will simply miss ‘surprisingly little’ societal change. The executive recently indicated that the target could be achieved sooner than expected. A technical employee of the company indicated that the publication of the company’s o1 reasoning model constitutes AGI. He admitted that while he does not exceed the cognitive abilities of a human, he is better at most tasks than humans.
Microsoft was seemingly left out, but the company’s AI CEO, Mustafa Suleyman, recently spoke with The Verge editor-in-chief Nilay Patelto discuss the rapid development of AI, including the AGI benchmark and whether this is achievable with current hardware.
For context, Mustafa Suleyman joined Microsoft earlier this year to lead the company’s operations across multiple products, including Copilot, Bing, Edge and other consumer AI products.
Microsoft AI CEO talks AGI, the OpenAI partnership and everything else
Early on, The Verge’s Patel asked Microsoft’s AI CEO if AGI could be achieved with current hardware. “I don’t think it can be done on (Nvidia) GB200s,” Mustafa Suleyman replied. “I think it will be plausible at some point in the next two to five generations.”
The question might focus on OpenAI CEO Sam Altman’s claims that AGi is feasible with current hardware. However, Altman indicated that while the benchmark requires new and sophisticated hardware, “you will be satisfied with a new device“.
The Microsoft AI CEO cast a wide net, indicating that the coveted AGI benchmark could be achieved in the next five to seven years with next-generation AI chips. The CEO was apparently walking on eggshells and categorically did not give a specific timeline for the AGI moment:
“I think it’s most likely in two years, but I think in the next five to seven years.”
Although Suleyman seems optimistic that AI systems can be trained without much prior solicitation, he is skeptical that getting things done in robotics will be a daunting task.
He added that this would not necessarily constitute an AGI or a singularity. Microsoft AI CEO defines AGI as “a general-purpose learning system that can work well in all human-level training environments.”
The executive said AGI has become so dramatized that the specific capabilities of the system are being overlooked. Suleyman indicated that he is much more interested in “creating AI companions, making them so that they are useful to you as a human, that they work for you as a human, that they are by your side, in your corner and in your team”.
Suleyman’s feelings are not surprising after Microsoft’s recent major update overhauling the Copilot user experience. The update was met with mixed emotions. Most users of the tool cited a degraded user experience coupled with poor quality responses.
Although Microsoft has yet to respond to comments, Microsoft’s AI CEO shared his vision for Copilot, indicating that it will eventually evolve into a virtual companion who can become a friend and foster meaningful, long-lasting relationships with users. Based on Copilot reviews on the App Store, users have already started to spot subtle hints of the platform’s evolution into a companion, and they’re not too happy about it. “He tries to be my friend when I need him as a tool,” said one concerned user.
The executive also discussed Microsoft’s multibillion-dollar partnership with OpenAIwhich has sparked backlash from Microsoft insiders, who say it has morphed into “a glorified hot start IT department” Despite the complex controversy over the tech bromance, Suleyman believes it could be “one of the most successful partnerships in computing history.”
Suleyman described Microsoft’s partnership with OpenAI as mutually beneficial. The former has access to advanced AI models and intellectual property (IP), while the latter has access to computing power and funding.
As you may know, Microsoft is expected to end its partnership with OpenAI once the latter achieves the coveted AGI benchmark. Asked about Microsoft’s backup plan when/if OpenAI achieves this impressive feat, AI CEO Mustafa Suleyman indicated that the definition of AGI is unclear.
Interestingly, a new report suggests that OpenAI is in discussions for remove the clause that cancels its partnership with Microsoft after reaching the AGI moment. The move could be designed to secure future funding and investment from the Redmond giant.
He also took the opportunity to tout Microsoft by talking about its capable and experienced AI team, including Karén Simonyan, who led the deep learning scaling team at DeepMind, leading to significant breakthroughs.
So we have an exceptional team and we will make sure that no matter what happens, we will be able to train the best models in the world.
Mustafa Suleyman, CEO of Microsoft AI
The Verge’s wide-ranging interview with Microsoft AI CEO Mustafa Suleyman covers many topics, including the company’s shift to security, which is now among its top priorities following a cascade of security breaches .
Suleyman also spoke about Microsoft CEO Satya Nadella’s leadership style, which focuses on accountability. He also compared Microsoft’s business model to Google’s, citing revenue and P&L discipline.