As another fork-in-the-road-esque time seems to be arriving for artificial intelligence, tech leaders mull over their next steps forward in 2025 with regard to user adoption, navigating a possible slowdown in pace of progress, and the matters of AGI.
Google CEO Sundar Pichai underlined how crucial 2025 will be at an internal meeting held recently. Admitting that the unwieldy size of the company had meant it was playing catch-up in artificial intelligence, he added that they had to work faster. “I think it is really important that we internalize the urgency of this moment and need to move faster as a company. The stakes are high. These are disruptive moments. In 2025, we need to be relentlessly focused on unlocking the benefits of this technology and solving real user problems,” he noted.
Most of this forward momentum was dependent on their Gemini models, where most of the company’s focus would lie. Pichai said that while the Gemini app could soon be reaching half a billion users, they still had “some work to do in 2025 and close the gap and establish a leadership position there as well.” To push adoption among users and “scale Gemini on the consumer side would be their biggest focus next year,” Pichai said. Still, concerned employees expressed doubts that OpenAI’s ChatGPT was slowly becoming “synonymous with AI the same way Google is with search.” Pichai assuaged these fears, saying that historically it has been proven that coming first isn’t always necessary, executing a product better than everyone else was. “I think that’s what 2025 is all about,” he noted.
Contrary to Pichai view that a slowdown is imminent in AI advancements, OpenAI CEO Sam Altman believes next year will not be another instance of tech hitting a wall. A month ago, when he sat down with Garry Tan, CEO of startup incubator, Y Combinator, for a video interview, Altman said, “AGI? Yeah, excited for that.” Artificial General Intelligence, is essentially the point at which a machine starts working like a human - it is able to not just generate text, but infer and then continue learning as it goes. Altman, however, has recently tempered expectations for what people expect AGI to look like.
Earlier this month, in an interview with The New York Times, he said, “My guess is we will hit AGI sooner than most people in the world think and it will matter much less.” He added that the economic disruption due to AGI would take some time to manifest in the world. “There’s a lot of inertia in society. So, in the first couple of years, maybe not that much changes and then maybe a lot of changes,” he noted. There’s also a peculiar clause between OpenAI and Microsoft around the AGI goal post. A new report by The Information said that if an AI system can generate at least $100 billion in profits, it would be considered as AGI, without specifying any other technical indicator. Interestingly, a previous report had revealed that once AGI is reached, Microsoft will not have access to OpenAI’s AI models. This could explain why Altman is rushing towards it. In the same interview, he said that “Artificial Super Intelligence” would be way more significant - which is a state where machines are superior to human intelligence. Meanwhile, both OpenAI’s new o3 and a new o3 mini model are currently in testing and will be made available soon next year.
In a podcast appearance two weeks ago, Microsoft CEO Satya Nadella spoke on several contentious topics. He believes next year will be very competitive despite the company’s close ties with the market leader OpenAI. However, he shared that it wouldn’t be a case of “winner-takes-all,” as gains from AI were going to become more diverse. Nadella also conceded that while Google did have its omnipresent search browser and a widespread mobile distribution network as an advantage, he was looking forward to regain lost ground with CoPilot and Edge. “This is the best news for Microsoft shareholders - that we lost so badly that we can now contest it and win back some share,” he said.
Nadella is banking on AI agents and has predicted that the business applications exist could “collapse” in the agentic AI era. This could render SaaS solutions much less important he said.
The CEO of Meta had boasted back in October that the next major AI model to be released - the Llama 4 - was being trained on a cluster of GPUs “bigger than anything else that’s been reported.” The AI model is reportedly scheduled for an early 2025 release. Zuckerberg has also been repeating the refrain that Meta AI, based on the Llama models, was well on its way to becoming the “most used AI assistant in the world.” Recently, this month, he noted that Meta AI “now has nearly 600 million monthly active users.” The company achieved these numbers presumably because of the massive user base it has amassed from its social applications including Instagram, Facebook, WhatsApp and Messenger. The AI assistant has been integrated in the search bar for all these apps pushing it to the forefront so AI became quickly accessible to consumers. In November, Meta hired Clara Shih, the former CEO of AI at Salesforce to lead a new Business AI group at the company indicating that they could be taking a stab at enterprise technology next.
During an analyst call in October, Zuckerberg described the road ahead. “It’s clear that there are a lot of new opportunities to use new AI advances to accelerate our core business that should have strong ROI over the next few years, so I think we should invest more there,” he stated. But he remained bullish on the technology saying it underpinned a lot of other areas in the company including its significant online advertising business and devices like the Ray-Ban smart glasses and the buzzy Orion headset which Zuckerberg called potentially the “next computing platform.”
Apple CEO Tim Cook recently defended the company’s AI strategy during an interview saying they weren’t lagging behind. The hardware giant’s philosophy had always been to “innovate in such a way that things would be personal and private.” He also dispelled rumours around them charging a subscription price for Apple Intelligence. “We never talked about charging for it. We view it sort of like multitouch, which enabled the smartphone revolution and the modern tablet,” he said. When asked about a possibility of an open ecosystem with third-party services for a more personalised AI system, Cook said that privacy must be prioritised. Although he wasn’t entirely against it, he said that the company wouldn’t accept a “trade-off between great privacy and great intelligence.”
While he did admit to having discussions around AGI, Cook said that at the moment they were focused on “delivering the technology to people and change their lives.” Notably, he revealed that Apple was looking to make inroads into consumer tools for medical technology via the Apple Watch and AirPods. “I’m not going to announce anything today. But we have research going on. We’re pouring all of ourselves in here, and we work on things that are years in the making. We were working on hearing a long time before we got it dialled in to where we felt comfortable shipping it,” he stated. AI was a key component here as Cook hinted at using the technology to analyse biometric data in real time for users.
Earlier in October, Amazon-backed Anthropic announced that it had built an AI system that could automate a range of tasks for coders i.e., an agent. But the AI firm is nonetheless planning to tread with caution when it comes to releasing AI agents in the wild. In an interview in December, co-founder and CEO, Dario Amodei called AI agents an “early product with low levels of reliability.” He warned users against trusting it with critical tasks. However, Amodei expects that there will be “a lot of progress made towards that by 2025.” While he said there will be products built around AI agents in 2025, “there will still always be tasks that you don’t quite trust an AI to do on its own.” Amodei said that he intended to spend most of his time next year with a couple of things in mind - firstly, to peek inside the large language models and study how they work and secondly, be involved in AI applications around biology. Amodei had done his postdoctoral scholarship from Stanford University School of Medicine but said he moved to AI later because biological problems had seemed “almost beyond human scale and comprehension because of how complex they were.” He revealed that Anthropic was already working with pharmaceutical companies and biotech start-ups. “I hope we start in 2025 to really work on the more blue-sky, long-term ambitious version of that — both with companies and with researchers and academics,” he added.
Published - December 31, 2024 01:53 pm IST
