Altman predicts that we are only 1,000 days from AGI, Hollywood goes all in for AI, Microsoft presses the nuclear button
Thought of the week: Sam Altman predicts super intelligence within 1,000 days!
Sam Altman just confirmed OpenAI's o1 model (codenamed Strawberry) has reached Level 2 on their 5-tier AI achievement scale. To recap, his five levels are:
Chatbots (current widely-available AI)
Reasoners (where the latest version of ChatGPT - o1/Strawberry - sits)
Agents (bots doing stuff for us autonomously - more about this below)
Innovators (capable of independent scientific discoveries - e.g. a cure for cancer)
Full organizations (Skynet ;-))
First of all we need to be cynical about Altman’s setting of the agenda. By defining ‘the’ path to AGI (artificial general intelligence) he’s clearly acting like an evangelical priest encouraging his flock to give donations to get them to the promised land (and Altman, having just raised a further $6billion, is very, very good at this!)
But cynicism aside, the Strawberry release is a step change up the ladder to Level 2. Taking medical diagnosis as an example, a Level 2 AI system should now be able to analyse a massive patient dataset, cross-reference symptoms with medical literature, and suggest a likely diagnosis, like identifying early-stage cancer. So far so useful, but, after presenting its findings, it stops there. It can’t independently decide to schedule tests, adjust treatments, or notify the patient—it still needs (annoyingly for AI proponents) a human to act on its conclusions and this is when a Level 3 system comes to the rescue.
A Level 3 AI wouldn’t just identify early-stage cancer; it could also order follow-up tests, notify the patient’s healthcare provider of the result, and even suggest a treatment plan—all without human oversight. In essence, Level 3 AI systems close the loop between analysis and action, shifting from being a tool for insights to a partner in execution.
Add this to what I call the holy grail of AI - systems that are:
contextual (they know you better than you know yourself)
conversational (forget that typing nonsense)
constant (with you 24/7)
and we’ll have agents that fully replace humans in many fields. Data analysts, accountants, project managers, management consultants…the list goes on.
If Altman’s timeline is even remotely accurate, and in fairness he’s been pretty spot on so far, then we’re on track to have Level 3 AI by this time next year. This means we finally get to see how useful (and job-destroying) AI can be once it’s unchained from its tiny and cumbersome chatbox interface.
Hollywood goes all in for AI
The AI video generation company Runway just announced that it has signed a deal with Lionsgate, the film studio behind such classics as the Hunger Games, Twilight and the John Wick series. Runway will essentially have access to will create a new customized AI model based on Lionsgate’s extensive back catalogue. Similarly, Stability AI, the company behind the generative AI model Stable Diffusion, have signed up Titanic filmmaker James Cameron who sees the future of filmmaking as a fusion of generative AI and CGI.
Lionsgate Vice Chair Michael Burns is very clear that this means making cheaper movies by using AI tools to create new footage based on old stuff.
“Runway is a visionary, best-in-class partner who will help us utilize AI to develop cutting edge, capital efficient content creation opportunities,”
The operative words here are “capital efficient”. The direction of travel is only one way. AI being deployed to cut as many creative costs as possible and speed up time to market.
AI goes nuclear - Big Tech’s desperate search for power
I’m a terrible investor. I passed up the opportunity to buy Google shares back in 2004 and my most recent foray into ‘the market’ - buying shares in the chip giant AMD -has been about as successful as a [insert political incorrect analogy]. However, if I was a betting man I’d put my money on power companies. Why? Because of the insatiable need of Big Tech companies for power.
Bloomberg, for example, just reported that Microsoft has agreed a deal to restart one of Three Mile Island’s nuclear reactors😲 - the other was shut down in 1979 after a nuclear accident. What I find amazing is that no one seems to find it strange that Microsoft, an IT solutions provider has managed to get a decommissioned nuclear power station back up and running without any regulatory resistance.
Ironically, when I asked Claude AI (which has a knowledge cutoff of April 2024) to proof this section of my newsletter, even it was incredulous:
Another example is the Sam Altman-backed nuclear startup, Oklo, which is trying to get a permit for Idaho’s first nuclear power station.
All of this massive investment in infrastructure is being driven by Big Tech’s move fast and break things ideology. Sam Altman posted this on his personal website:
If we want to put AI into the hands of as many people as possible, we need to drive down the cost of compute and make it abundant (which requires lots of energy and chips). If we don’t build enough infrastructure, AI will be a very limited resource that wars get fought over and that becomes mostly a tool for rich people.
As always, nothing about the inherent peril of AI (job losses, privacy, lack of agency) just an extremely well-funded company hellbent on shaping the world through the lens of a hyper efficient, data-driven decision making system that obviates the need for humans.
AI video of the week
What we’re reading this week
Elon Musk’s brain chip firm Neuralink has received “breakthrough device” status for an implant aimed at helping blind patients regain their sight.
A project Analyzing Human Language Usage Shuts Down Because ‘Generative AI Has Polluted the Data’😲
LinkedIn Is Training AI on User Data Before Updating Its Terms of Service
AI pioneers called for international oversight to address potential catastrophic risks from rapidly advancing AI technology, warning it could soon surpass human control.
Great article about Scale, the Chinese company that deploys an army of global workers to keep AI on track.
Tools we’re playing with
Unfortunately not ChatGPT’s Advanced Voice Mode as it’s unavailable in the UK 😒
Strawberry - see my efforts at Creating a Chess Game with Chat GPT Strawberry 🎮 - Watch Video
That's all for this week. Subscribe for the latest innovations and developments with AI.
So you don't miss a thing, follow our Instagram and X pages for more creative content and insights into our work and what we do.
