Leo's amazing predictions

[Stick]

it doesnt need to be intelligent though. It can be dumb as a rock but as long as it gets the job done better and cheaper than the avg worker thats all thats needed for mass lay-offs.

[Natasha Tori Maru]

3 minutes ago, Stick said:

It can be dumb as a rock but as long as it gets the job done better and cheaper than the avg worker thats all thats needed for mass lay-offs.

From "dumb as a rock" --> job done is such a fucking massive logical and technological leap!

Where is that happening, what sort of timelines, who's job, how is it going to happen? 

None of this is mapped yet. 

So as far as I see, this is just fantasy thinking. Might happen. Might not. A fart in the wind.

Edited 3 hours ago by Natasha Tori Maru

[something_else]

15 minutes ago, Leo Gura said:

But AI is not more intelligent than even a dumb human, even a teen.

AI is less intelligent than 100% of humans.

In the ways that matter for achieving results in white collar jobs (largely text and data driven), it is almost certainly more capable than the average person within the domain of those careers.

Programming is a good example because Claude is a much better programmer than your average junior/mid-level developer, hence why companies are hiring much fewer junior/mid level developers. It's more efficient to scale out AI agents and let seniors guide them.

As another example, Claude can solve university level quantum mechanics papers with a higher success rate than an average student taking those classes.

You'll also get a better philosophical discussion out of an AI agent than probably 99.9% of humans.

Are these the be all end all of intelligence, or is it AGI? I agree, no. But it certainly is a kind of intelligence, and it has more of it than your average human in many fields. And enough of it that it isn't just going to go away.

Quote

If you want to claim AI is more intelligent than large % of the population - you better come to the party with what you base that on, and your reasoning... 

Perhaps it isn't correct to say that AI agents are more intelligent than most humans, but they are certainly more capable in a fairly wide variety of fields. And ultimately this is where most value comes from when using these tools. They produce incredible results that no human could in the same time frame.

[something_else]

28 minutes ago, Natasha Tori Maru said:

From "dumb as a rock" --> job done is such a fucking massive logical and technological leap!

Where is that happening, what sort of timelines, who's job, how is it going to happen? 

None of this is mapped yet. 

So as far as I see, this is just fantasy thinking. Might happen. Might not. A fart in the wind.

I don't know if you've ever seen proper agentic AI working. Most of the chatbots you see embedded in Google, Gemini, ChatGPT etc. are just that: fancy chatbots that are good at writing text.

Once you put those models in an agent harness, which allows the LLM to loop, talk to itself, think, use tools, operate autonomously for long periods of time etc. they are incredibly competent. A lot of people who aren't in tech haven't seen these agents work. It's extraordinary.

You can give them a task (a single prompt of a few lines of text for what you need them to do) and the process it will follow after this (fully autonomously) could look something like:

• talking to itself to come up with a very rough overall approach to take
• brainstorm ideas
• connect to tens of external services to pull in the information it needs
• research what the best practices / standards are for this task
• fire of 5 subagents primed with different skillsets to go off and do some more research in different areas
• spawn 5 more subagents to adversarially critique or summarise the output of those first subagents
• compile all of those findings + brainstorming down into a plan
• execute that plan via connecting to external services or modifying text/code/data in whatever mediums it needs to to achieve the end result
• Once they have an end result, they'll test it extensively and check that what they implemented works
• Perhaps it will spawn some more subagents to adversarially critique and verify the end result until it fixes every problem it has found with its own implementation
• Publish/push/commit/save the end result of that task for you with a tidy summary

And all of this takes like 10-30 minutes. To finish a task that may have taken a human a week. It isn't perfect and it will make mistakes, which is why it typically works best when done in combination with someone who is competent in the field. But in the past you would have one senior working with 3 juniors and a mid-level, now you may just have a senior working with one AI agent.

Edited 2 hours ago by something_else

[Natasha Tori Maru]

@something_else but are they at the stage they can do that, without needing a lot of oversight? As much resources to oversee that might have gone toward doing the tasks? It may finish quick, but how much time and resources go to checking it all?

Agentic AI being able to run autonomously I am interested in, but I haven't seen anything to indicate it is anywhere near that level yet. Or seen a decent trajectory for when. 

I don't see a list of pure capabilities outside of the context of real world use as supplying much to forward the argument it will replace a human intelligence without the oversight.

I won't buy any sort of future predictions as being close to accurate until I see AI being able to coherently interact with 3D reality and space and operate in conjunction with robotics. 

I'd love to see a huge list of tangeable examples of real application. Not nebulous capabilites. Also sort of interested in what you work with? What sort of agentic AI do you use and how is it giving you results that are shortcutting resources? Or is this from external sources?

Edited 2 hours ago by Natasha Tori Maru

[Natasha Tori Maru]

You can still see there is a massive leap from current AI ---> replacing humans (job wise/ intellegince). Huge. 

[Alex M]

On 09/07/2026 at 11:28 AM, Leo Gura said:

If AI can replace jobs so easily, why don't we start first with CEO jobs?

Why don't we replace Elon Musk with Grok?

If AI cannot replace Elon Musk, it cannot replace a programmer.

These tech CEOs would themselves be fucked if AI worked like their fantasies believe.

They are trying to sell us a monorail. I am not buying it.

Dam it, I wanted to get a job as the monorail conductor. But in all seriousness, the levels of madness and overhype parallel those of the early dot-com bubble. Investors are just throwing money at it. 

[NewKidOnTheBlock]

Let's be real here, most of you here are professional commenters, you know what I mean. We'll see how things will pan out, and I'd rather follow opinions of researchers directly working with this technology

[something_else]

24 minutes ago, Natasha Tori Maru said:

but are they at the stage they can do that, without needing a lot of oversight? As much resources to oversee that might have gone toward doing the tasks? It may finish quick, but how much time and resources go to checking it all?

This is a good question. In short, yes, they can. They need oversight but not nearly as much oversight effort as it would have taken to do the task yourself.

I can give a very concrete example of how this works for my own field/job which is software engineering. This is what has happened at the company where I work:

Before AI adoption

Previously we were a team of 8 developers, which was split into two squads of 4. Each squad works on 1 project at a time, so the entire company is working on 2 projects at a time. 

Those projects would typically aim to take anywhere from 2 weeks to 2 months to complete and would be for small to medium sized features. Occasionally we would have 'large' features which we would set aside 3-6 months for a team to implement.

After AI adoption

2 devs left and the company opted not to replace them, so we are a team of 6 now. We have restructured into 'pods' where each of our 6 devs works on a single feature at a time with the help of AI, a human designer + human product manager.

The AI agent writes the vast majority of the code and the developer guides it with some oversight and prompting. Usually one other dev will review the work before it is released, but this is probably an average of 30 minutes worth of effort of review per dev per day.

We no longer distinguish much between feature sizes. Almost any feature we could want to implement can be implemented within 1, maybe 2 months.

We are shipping 6 medium/large features in the same amount of time we would previously have shipped 2 medium features, all while having 2 less devs employed.

Edited 2 hours ago by something_else

[Vynce]

If the investment bubble assumes AGI and AGI assumes consciousness to arise through semi-conductors circuits, it is chopped either way. 
 

A good example how bad beliefs and epistemic ignorance lead to trillions wasted into pipe dreams. 

If they knew what consciousness really is investments would not exceed 20% of current spending.

[Girzo]

AI is great for some boring tasks. For example, rewriting academic paper drafts into a completed work. A human has to make a sensible draft and good points, but it saves a lot of time. I can write whatever I want, no care about grammar and errors, just put insights on paper and then have a few rounds with Claude to make it into a flowy, finished text. At the end you have to go through each and every sentence to see if it hadn't garbled the meaning, but it still saves time.

Edited 48 minutes ago by Girzo