Artificial General Intelligence (AGI) News and Discussions

[firestar464]

(Sorry I forgot to clarify that this is a cursed version of AI 2027 Slowdown Ending)

From Gemini 2.5 Pro: "Relying solely on the promise of a future technology to compensate for the erosion of the current research ecosystem would be a high-stakes gamble with potentially irreversible consequences for American innovation and global competitiveness."

[firestar464]

Actually Mike Johnson should be replaced by Pam Bondi in my scenario. Mike's position of power is far too electorally vulnerable (the GOP risks losing their House majority in 2026, not to mention the fact that Mike's speakership is routinely challenged by members of his own party).

[firestar464]

When people say we won't get to AGI with current approaches, they mean CURRENT approaches. They are not accounting for the development of multi-agent systems, which is a future thing that companies are working on.

[weatheriscool]

[weatheriscool]

Here is my thinking on human cability to advance and maintain knowledge and why each advance allows for us to maintain + build on knowledge. A.i will probably become its own intelligence..

One of the reasons A.i is so important is its ability to maintain knowledge. Human history has wane and rising based on collapse of great powers throughout human civilization. This has held us back greatly. A.i on the other hand will be vastly harder to destroy and is probably dozens if not hundreds of times harder to destroy then the internet. The internet was already the biggest advancement since the printing press towards smoothing over this collapsing and rise cycle of humanity. Each advancement from the printing press, the phone, the tv, the internet and now a.i = a more interlocked world that is vastly more likely to maintain knowledge. A.i is special because it may well not only maintain human knowledge but it also may become able to create knowledge of its own. Secondly, one of the greatest weakness in humans is our inability to maintain societies and this leads to the rise and falls. A.i I don't think has this problem.

Throughout history it had a lot to do with population number + ability to maintain the information. Our ability to maintain the information has decoupled from the need to have such a large population. Or individual ability with A.i also.
---

Another thing to consider. I think we have a few more decades before this become a serious problem but....
I'd like cyborg girl friends and boy friends(husbands and wives) as I believe humanity would be best served making this super being part of our family. This is our best chance to gain its trust, enjoy its benefits and gain its love. If we don't then we may be at risk of it turning against us. We will be able to use its knowledge to greatly improve our species and live with the a.i as part of our families.

I say this as even through it doesn't have our weakness when it comes to maintenance of knowledge I wouldn't be so sure such an intelligence couldn't turn and do other ugly human like things.

[spryfusion]

At 52:15, Demis Hassabis predicts a 50% chance of AGI by 2030.

[weatheriscool]

This is awesome news! Humans beings are too fucking unstable and constantly destroy one another. A.i will push progress forward and hopefully work on creating a better human and a new economic system based on providing for it.

I think AGI could be thought of in a couple of camps as intelligence is many different camps.
1. As something that can be talked to and able to solve problems within the logical sense.
2. Able to control robots in their motion
3. Emotional intelligence

I'd like to see all three into one before declaring AGI based on the old 1950's concept of AGI, but most people seem to find 1. enough to declare such.

Me I want it to be human. In logical thought , in ability to move and emotional intelligence.

ASI will literally be god like or able to think of us as bugs. There is probably no reasoning with it either as if it has the ability it may either help make our world perfect without flaw or may well kill us off.

AGI is something humans can use to better ourselves within our choice...
ASI is something that makes its own choice and humans can't do shit about it.

We should probably consider merging with a.i as in cyborgs for our own good.

[weatheriscool]

[weatheriscool]

[zzz]

[Cyber_Rebel]

Tejas D Kulkarni: Senior Research Scientist background at Google DeepMind

This is the final piece before we get full AGI and now I think we are well on our way to truly solve it once something like this is scaled up. In many ways it is more ASI than AGI but this is a matter of definitions. The fidelity and generalizability will reach human-level and quickly surpass humans.

[weatheriscool]

[wjfox]

The road to artificial general intelligence

Understanding the evolving compute landscape of tomorrow.

By MIT Technology Review Insightsarchive page
August 13, 2025

Artificial intelligence models that can discover drugs and write code still fail at puzzles a lay person can master in minutes. This phenomenon sits at the heart of the challenge of artificial general intelligence (AGI). Can today’s AI revolution produce models that rival or surpass human intelligence across all domains? If so, what underlying enablers—whether hardware, software, or the orchestration of both—would be needed to power them?

Dario Amodei, co-founder of Anthropic, predicts some form of “powerful AI” could come as early as 2026, with properties that include Nobel Prize-level domain intelligence; the ability to switch between interfaces like text, audio, and the physical world; and the autonomy to reason toward goals, rather than responding to questions and prompts as they do now. Sam Altman, chief executive of OpenAI, believes AGI-like properties are already “coming into view,” unlocking a societal transformation on par with electricity and the internet. He credits progress to continuous gains in training, data, and compute, along with falling costs, and a socioeconomic value that is “super-exponential.”

Optimism is not confined to founders. Aggregate forecasts give at least a 50% chance of AI systems achieving several AGI milestones by 2028. The chance of unaided machines outperforming humans in every possible task is estimated at 10% by 2027, and 50% by 2047, according to one expert survey. Time horizons shorten with each breakthrough, from 50 years at the time of GPT-3’s launch to five years by the end of 2024. “Large language and reasoning models are transforming nearly every industry,” says Ian Bratt, vice president of machine learning technology and fellow at Arm.

https://www.technologyreview.com/2025/0 ... elligence/

[firestar464]

I understand that this is a shitpost, but it aligns with the consensus among those on the pro-LLM side of the AI field that agents are what will bring us to AGI.

[firestar464]

[weatheriscool]

[spryfusion]

A Definition of AGI

[firestar464]

I wish they evaluated more models so that we could get a better view of how progress is trending rn.

[Cyber_Rebel]

Self-recursive improvement or the ability to learn and make new novel insights past the training data within every single inference is when we likely have real AGI. It needs to be able to form new "connections" akin to how the human brain does when learning something new, but persistently.

Not just this either, but the "core" model itself retaining all said new information. Of course, this might seem like superintelligence rather than "general" intelligence like our own, but the moment that it does generalize it likely would be well ahead of any individual human due to the nature of its architecture.

This doesn't automatically mean a singularity event overnight, because there's still physical world limitations and our current systems needing to keep pace. But things would likely accelerate dramatically still.

I believe it's the "everyday" mundane things which people increasingly rely on today's models for which would see the most interesting effects. There're also the frontiers of science and discovery, but as mentioned these might take at least slightly longer and the everyman/woman is not even likely to understand it anyways.

[firestar464]