Learning Library

← Back to Library

Sam Altman's 2025 AGI Outlook

Key Points

  • Sam Altman’s New Year’s Reflections predict the arrival of artificial general intelligence (AGI) in 2025, specifically in the form of AI agents that act as colleagues in tools like Slack.
  • These “AI coworkers” are expected to perform enough work to be billed at roughly ten percent of an equivalent employee’s salary, but they will still require human oversight and cannot replace entire organizations.
  • Altman foresees a short‑term shift where companies hire these agents for specific tasks, causing minimal disruption to the broader labor market.
  • He also signals a strategic pivot for OpenAI: half the business will continue offering AGI‑powered products (e.g., ChatGPT), while the other half will refocus on core research aimed at achieving superintelligence.
  • The ultimate goal of that superintelligence, according to Altman, is to accelerate scientific breakthroughs—potentially solving hard problems like aging, cancer, and other unsolved diseases.

Full Transcript

# Sam Altman's 2025 AGI Outlook **Source:** [https://www.youtube.com/watch?v=7fJSM4G9Efg](https://www.youtube.com/watch?v=7fJSM4G9Efg) **Duration:** 00:04:29 ## Summary - Sam Altman’s New Year’s Reflections predict the arrival of artificial general intelligence (AGI) in 2025, specifically in the form of AI agents that act as colleagues in tools like Slack. - These “AI coworkers” are expected to perform enough work to be billed at roughly ten percent of an equivalent employee’s salary, but they will still require human oversight and cannot replace entire organizations. - Altman foresees a short‑term shift where companies hire these agents for specific tasks, causing minimal disruption to the broader labor market. - He also signals a strategic pivot for OpenAI: half the business will continue offering AGI‑powered products (e.g., ChatGPT), while the other half will refocus on core research aimed at achieving superintelligence. - The ultimate goal of that superintelligence, according to Altman, is to accelerate scientific breakthroughs—potentially solving hard problems like aging, cancer, and other unsolved diseases. ## Sections - [00:00:00](https://www.youtube.com/watch?v=7fJSM4G9Efg&t=0s) **Sam Altman's 2025 AGI Forecast** - Sam Altman predicts artificial general intelligence will appear in 2025 as AI agents that function as workplace colleagues—costing roughly 10% of an employee’s salary, causing limited short‑term disruption and prompting a shift in business focus. ## Full Transcript
0:00you know when normal people write their 0:01New Year's resolutions and their 0:03Reflections nobody pays attention but 0:05when Sam Altman does founder of open AI 0:07everybody does so he wrote his new 0:09year's Reflections I read through them 0:11for you long story short Sam sees 0:15artificial general intelligence this 0:17year in 2025 and specifically he named 0:20The Format he thinks it will take which 0:22I think is important he hasn't said that 0:24publicly before he says it's going to 0:27look like AI AG agents that are 0:31colleagues in slack AI agents that are 0:34colleagues at work that you can partner 0:36with to get jobs 0:37done that's it that's that's what he 0:40defined as important enough and he 0:42thinks that they will do good enough 0:44work that companies will pay for them 0:47now if I were to guess he did not say 0:49this but if I were to guess I would say 0:50typically that's going to get priced in 0:52about 10% of the equivalent employee 0:55salary and before everyone says well 0:57that's the end of jobs as we know it I 0:59don't think so because these agents by 1:03definition can't do the work of an 1:04organization yet and so they do need 1:07humans around them above them below them 1:09Etc and so I think what we're going to 1:12see is just some new AI colleagues and 1:15that's what it's going to look like and 1:16it will actually be fairly minimal 1:18disruption in the short term at the same 1:21time Sam mman shared something else 1:23that's interesting he thinks that the 1:25focus of the entire business is going to 1:28shift from on the one hand running these 1:30sort of generally intelligent agents as 1:33a product maybe running Chad GPT as a 1:35product with one half of the business 1:38and then the other half is going to go 1:39back to what open AI has traditionally 1:41focused on which is research and 1:44specifically research towards super 1:47intelligence and this is something that 1:49over the holiday break various open AI 1:52Associated Affiliated working at kind of 1:55people were saying we see something we 1:57think that super intelligence is near 1:59this is sort of Sam's way of saying he 2:00sees a path to Super intelligence and 2:03typically super intelligence has been 2:05defined as an AI that can do the work of 2:08an entire organization not just one 2:10person but interestingly enough when Sam 2:14talked about it in his Reflections that 2:16wasn't the definition he focused on he 2:20might not disagree with it but what he 2:21focused on was the power of super 2:24intelligence to accelerate scientific 2:27research and I think he's really excited 2:29because he sees some of our biggest 2:31hardest problems as a species as 2:33something that super intelligence could 2:35help us solve so the classic one is 2:38aging what if super intelligence could 2:39find a solution for aging and then we 2:42wouldn't have to age and 2:43die right like that's as old as like 2:46Humanity's quest to to avoid death right 2:48like that's not new uh but Sam is 2:50thinking that super intelligence might 2:52be able to help with that there's a 2:54bunch of other ones right like uh long 2:56unsolved diseases quote unquote the cure 2:59for cancer sir um and so I think what 3:02Sam is interested in and where he hinted 3:04that open AI is going is that as they 3:06get closer to Super intelligence they're 3:09going to Pivot more into a scientific 3:11research lab and partner more with 3:13scientific institutions I think you saw 3:15a hint of that in the holiday 3:17announcements when they tipped their cap 3:19a few times on a few of those 12 days of 3:22open AI to Scientific institutions into 3:24how they help researchers I think they 3:27want to do that more um I think they 3:29want to build closer partner closer 3:30partnership links I think they want to 3:32do some scientific research themselves 3:35and I think that's something that Sam 3:36sees as a long-term Legacy for the 3:38company if he can solve some of these 3:40big problems with super intelligence he 3:41will feel like the work that he did had 3:44some meaning to it and that's something 3:46that in in our small way that's that's 3:48what we all want right uh and it's very 3:51appropriate for a New Year's reflection 3:52so there you go uh Sam thinks that 3:54general intelligence is coming this year 3:57and Sam also thinks that uh super 4:01intelligence is the next big project 4:02that open AI has he did not put a 4:04timeline on that one by the way so we'll 4:06see that by the way is my first 4:09prediction from 2025 that I got right I 4:10was very pleased with myself so I I I 4:12was guessing that we were going to get 4:14uh agents uh like in slack collaborating 4:17with our with us as colleagues um so I 4:19need to take one all the others I could 4:21still get wrong but I'll take one uh I I 4:23put that up on my substack it was fun to 4:25see that one come right in the first few 4:26days of the year all right cheers