It sure is. That’s why I mentioned a jellyfish in the very first comment. Past 500M years it became… a jellyfish.
Progress is inevitable. That means sooner or later the evolution would have come to AGI. It does not mean the previous step is AgenticAI.
Think many would disagree with you Oliver…
While Artificial General Intelligence (AGI)—AI systems with human-level capability across all domains—is generally considered theoretically possible, consensus on its timeline is split between a few years to several decades.
In a nutshell
- Four key factors are driving AI progress: larger base models, teaching models to reason, increasing models’ thinking time, and building agent scaffolding for multi-step tasks. These are underpinned by increasing computational power to run and train AI systems, as well as increasing human capital going into algorithmic research.
- All of these drivers are set to continue until 2028 and perhaps until 2032.
- This means we should expect major further gains in AI performance. We don’t know how large they’ll be, but extrapolating recent trends on benchmarks suggests we’ll reach systems with beyond-human performance in coding and scientific reasoning, and that can autonomously complete multi-week projects.
- Whether we call these systems ‘AGI’ or not, they could be sufficient to enable AI research itself, robotics, the technology industry, and scientific research to accelerate, leading to transformative impacts.
- Alternatively, AI might fail to overcome issues with ill-defined, high-context work over long time horizons and remain a tool (even if much improved compared to today).
- Increasing AI performance requires exponential growth in investment and the research workforce. At current rates, we will likely start to reach bottlenecks around 2030. Simplifying a bit, that means we’ll likely either reach AGI by around 2030 or see progress slow significantly. Hybrid scenarios are also possible, but the next five years seem especially crucial.
The godfather of AI himself:
Thats a lot of text I don’t really feel like sifting through, but to me that graph looks about the same as:
I make a substantial living off of the benefits of AI assisted development and I spend a lot of time on the bleeding edge and if anything every new release makes me sure we are firmly in a “smartness” plateau. Only the RLHF and harness engineering is now progressing meaningfully.
Well if we get an AGI or a Super Intelligence, we are toast. Programmers become unnecessary in one year if we are lucky. If that SI has enough computing resources, it’ll achieve a full convergence – it will find 4-5 programming languages that allow it to progress quickly (like statically strongly typed and with super-fast startup time… so, Rust / Zig / D, basically), then it will start striving to use just one PL, then it will fix the one that it feels has the least blockers or problems, it will make a mega-uber compiler and a runtime, will start working only with it, then will drill deeper and will hyper-optimize the machine code, and then it will use the extra processing speed to accelerate its own development even more.
From then on we the humans are practically redundant and just a drag.
We should start working on positronic brains with the 3 laws of robotics hardcoded inside the hardware. 
I like to think we’d be able to create some sort of QSI (quarantined SI) and it’ll hopefully help us with things like mastering genetics. Perhaps leading to our own evolution, transhumanism.
Until that time… surprised nobody has said that AI would love to use a language like Elixir or Erlang, are we forgetting Robert’s comment:
Shouldn’t the blue “SI” piece be a triangle?
If super intelligence is defined as being intelligent to a point that we cannot comprehend it’s reasoning then I think you would be better served reading theology. We will have moved into territory that is not covered by any human language, let alone Elixir.
Whereof one cannot speak, thereof one must be silent
-Ludwig Wittgenstein
Personally I worry more about the economical collapse once the LLM claims are exposed to be hot air.
and the discussion around it on Lobste.rs is pretty good:
Yep.
I also worry about who’s (ultimately) going to foot the pattern-matching-search-engines-massive-private-credit-defaults bailout bill, but I’m sure the act itself will be named using a misleading and cynical acronym (e.g. PROSPERITY act, SOLIDARITY act or something sounding equally virtuous).
The buyout is definitely a big concern. Did this also happen with the dotcom bubble? It seems that the crypto miners are now also shifting to pander to the AI companies since they are the people with power connections and an existing datacenter. I wonder if they will go back into crypto mining when AI collapses. Could be a double whammy….
Prediction markets (maybe the next hype) use a lot less resources to provide gambling to the masses, so that hype will hopefully only make humans suffer from their actions. But if it is going to be quantum, then we will have another power hungry beast that will prop up the level of the oceans.
