OpenAI CEO Sam Altman believes that by 2026, artificial intelligence (AI) will be capable of helping businesses solve complex and important problems.
Speaking at Snowflake Summit 2025, he also teased that the company's AI models set to be released over the next year or two will be "quite breathtaking", with a lot of improvements across the board.
"Next year, we will be at a point where you can not only use a system to automate business processes and build new products and services, but you can say, 'I have this hugely important problem in my business. I will throw a ton of compute at it if you can solve it,' and the models can figure out things that teams of people on their own can't do," Altman said during a fireside chat with Snowflake CEO Sridhar Ramaswamy. The conversation was moderated by Sarah Guo, founder of venture capital firm Conviction.
Altman said that companies already experienced with such models will be well positioned for a future where they can ask AI systems to "think hard and redo their most critical project". "People who are ready for that will have another big step change next year. For instance, a chip design company can say, 'Go design me a better chip than I could have possibly had before'," he said.
These comments come after OpenAI reached an inflection point in how businesses were using its AI models sometime last year, according to Altman, leading to strong growth in its enterprise segment.
"As a general principle of technology, when things are changing quickly, the companies that have the fastest iteration speed, make the cost of mistakes the lowest, and the learning rate the highest, win," he said.
Ramaswamy echoed a similar sentiment, stating that curiosity is the "most overlooked thing" and it's important to keep experimenting.
"I think there's so much that we take for granted about how things used to work that's just not true anymore. OpenAI and Snowflake have made the cost of experimenting very low. You can run lots of little experiments, get value from it, and build on that strength" he said.
Evolution of AI agents
The OpenAI chief expects AI agents to soon help discover new knowledge and said they are already beginning to act like junior employees.
"You hear from companies that are building agents to automate most of their customer support, or their outbound sales and you hear people that talk about their job now is to assign work to a bunch of agents, look at the quality, figure out how it fits together, give feedback, and it sounds a lot like how they work with a team of still relatively junior employees" he said.
Altman said the work of AI agents is currently in the realm of repetitive cognitive work which can be automated at a low level on a short time horizon.
By next year, he expects these agents to help discover new knowledge in some limited cases or figure out solutions to business problems that are non-trivial.
"As that expands to longer time horizons and higher levels, at some point you can get an AI scientist, an AI agent that can go and discover new science, and that would be a significant moment in the world," Altman said.
In January, OpenAI launched its first AI agent, Operator, that can use its own browser to perform repetitive tasks for consumers. This includes filling out forms or expense reports, buying groceries, making restaurant reservations, booking a taxi ride, or making e-commerce purchases among others.
Initially available in the United States, the company extended it to ChatGPT Pro subscribers in India and several other markets like Australia, Brazil, Singapore, the United Kingdom, Canada, and South Korea in February.
In the same month, OpenAI also unveiled Deep Research, an AI agent capable of performing in-depth, multi-step research on the Internet and compiling reports on complex topics. The agent also enables people to connect files from third-party services such as Box, Dropbox, GitHub, Microsoft OneDrive, and SharePoint, for real-time analysis.
Last month, the company introduced a research preview of Codex, a software engineering agent that can work on many tasks in parallel. This includes writing features, answering questions about your codebase, fixing bugs, and proposing pull requests for review among others.
"Codex has been one of my "feel the AGI" moments. You watch this thing, give it a bunch of tasks, it goes and works in the background. it can also do these long horizon things" he said.
Altman further added "At some point, it'll be able to also watch your meetings if you want, look at your Slack, and read all your internal documents. Maybe, today it is like an intern that can work for a couple of hours, but at some point it'll be like an experienced software engineer that can work for days"
'How far AGI is doesn’t matter'
The OpenAI co-founder also argued that what AGI is and when it will arrive doesn't matter, instead the current rate of progress should sustain well beyond the next five years.
"To me, a system that can either autonomously discover new science or become such an incredible tool for people that our global rate of scientific discovery quadruples. That would satisfy any test of AGI I can imagine" Altman said.
Ramaswamy also called it as a philosophical question "I see these models as having incredible capabilities that anyone looking at what things will be like in 2030 would simply declare as AGI"
"I don't think the actual moment matters a whole lot...To me, it's the rate of progress that is truly astonishing. And I sincerely believe that many great things are going to come out of it" he said.
(The journalist is in San Francisco at the invitation of Snowflake)
Discover the latest Business News, Sensex, and Nifty updates. Obtain Personal Finance insights, tax queries, and expert opinions on Moneycontrol or download the Moneycontrol App to stay updated!
