Quantcast
Channel: Joab Jackson, Author at The New Stack
Viewing all articles
Browse latest Browse all 636

OpenAI’s Sam Altman: AI Is Now Ready for the Enterprise

$
0
0

Sam Altman

Sam Altman

“Just do it,” is the advice of Sam Altman for enterprise leaders wondering how to get their organizations started down the path of AI.

Of course, the OpenAI CEO has always been bullish about getting started on AI. But this is not the advice he would have given to enterprise execs even a year ago, he said, in a “fireside” discussion with Snowflake CEO Sridhar Ramaswamy for the kickoff of this year’s Snowflake Summit, being held this week in San Francisco.

Conviction Founder Sarah Guo moderated the discussion.

“Certainly, what you are seeing with enterprises and AI is that the people making the early bets and learning very quickly are doing much better than the people who are waiting to see how it’s all going to shake out,” he said.

“There’s never going to be one perfect moment when everything settles down,” agreed Ramaswamy.

But a year ago, Altman explained, AI was still too new for the enterprise. Last year, it was like the new intern in the office, a faster learner but still with novice capabilities of making embarrassing mistakes.

But this year, his ChatGPT and similar competing services are a bit more seasoned, more like handy junior engineers. They still get stuff wrong, but they can be trusted with routine tasks.

In fact, some have argued this is why we are already seeing a dearth of entry-level jobs across many sectors due to AI.

Now Is the Time for Enterprise

Altman said that a year ago, he advised enterprises not to plunge into AI-driven systems, and instead to just experiment with them. Enterprises need reliability and accountability above all, and that in 2024,  AI was “not ready for production use in most cases,” he admitted.

But AI has changed even in the last 12 months. OpenAI has seen a significant spike in enterprise usage, with customers being increasingly astonished at the things AI “can reliably do.”

“It does seem that in the last year, we’ve seen a real inflection point for the usability of these models,” he said.

And Large Language Models are very fast learners. Scary fast.

“The range of products is truly astonishing,” Ramaswamy agreed.

Within another year’s time, we will be able to trust LLMs not only to automate some business processes but trust them (with a great deal of supporting compute) to solve some of our most difficult, pressing problems, he predicted.

A chip company could have an LLM design a new processor. A biotech company will be able to have it work on some heretofore incurable disease.

“The models over the next year or two will be quite breathtaking,” Altman said. As a result, businesses will be able to do things that previously weren’t possible.

The Next Inflection Point

So, what will be the next set of emergent behaviors we’ll see, Guo asked.

The next inflection point, Altman predicted, is when AI agents will start autonomously discovering new information.

The platonic ideal of this would be a “very tiny model” that has “super-human reasoning abilities”  and can run ridiculously fast and have access to every tool possible.

Altman admits OpenAI is a long way from shipping such a model, but “directionally we’re headed there.”

Altman pointed out that only recently have models started doing abstract summarization of text well. Summarizing a large body of words within a few sentences can be a surprisingly difficult task for humans, and one that we only recently found out that LLMs do really well at.

Expect to see similar jumps in the years to come.

Guo then asked if Altman had 1000 times the compute power he had today, what would he do, and Altman said he would apply it to AI research, in order to build much better models.

Altman’s ambitions for AI are staggering, and given the fundraising he has to do to keep OpenAI progressing, perhaps they are even necessary. But others are foreseeing great jumps ahead as well.

A new report from venture capitalist Mary Meeker suggests that AI adoption is outpacing every other technology in human history.

The post OpenAI’s Sam Altman: AI Is Now Ready for the Enterprise appeared first on The New Stack.

Sam Altman now advises enterprise leaders to "just do it" when it comes to adopting AI, a significant shift from his more cautious stance a year ago.

Viewing all articles
Browse latest Browse all 636

Latest Images

Trending Articles



Latest Images