Leo's amazing predictions

[something_else]

7 hours ago, Leo Gura said:

 

@Leo Gura I respect the point this dude is trying to make, but he is presenting this as a duality of “to vibe code or not to vibe code” which is an (arguably intentional) misrepresentation of the issue.

The fact is that the majority of dev teams are incorporating powerful AI coding tools into their systems and workflows now. If done right most of the risks he warns about simply vanish.

As a concrete example, the company where I work has quadrupled our rate of shipping features in the past year. This is also despite 2 out of our 8 developers leaving. Those features are some of the most complex we’ve ever built, yet our error/bug rate has not increased substantially as a result of using AI coding agents to build them, in fact it has decreased.

Where I agree with him is that coding standards are extremely important. But as long as you define your coding/architecture standards to the AI model once it will largely follow them for the rest of time. When it doesn’t you can nudge it in the right direction.

The fact is that these coding agents are incredibly powerful and they are never going to go away even if they only ever stay at their current level of capability. They do not replace developers per se, but they allow companies to operate with a much smaller and more dynamic team of developers than they would have needed before, hence people’s deep concern for the software job market

Edited 1 hour ago by something_else

[Jodistrict]

The last time there was an AI bubble crash was cerca 1987.  Up to that time, the main paradigm was symbolic processing which culminated in expert systems.  But expert systems had a major flaw.  They couldn’t learn.  So after the rules got to a size of millions it was a nightmare and programmers had to manually update the databases.  In other words, expert systems couldn’t be scaled up.   

AI then gradually shifted to neural networks and the connectionist paradigm.  Neural networks have learning built into the system, and the rules are replaced by weights modified by data.  Deep learning and LLMs are built with mathematical advances on the neural networks.   In addition, the hardware technology available has significantly advanced by orders of magnitude.    

Thus, the last AI crash was due to a fatal flaw.  The current AI crash will be due to over investment as the technology requires billions of dollars of up front costs.  But the technology has already proven itself and it is scalable.   So the nature and duration of the crash will be different.

 

Edited 1 hour ago by Jodistrict

[Butters]

How can you debate the usefulness of AI when it literally changed warfare? Fully autonomous drones in the Ukraine and Iran wars is now the standard. So"it's just a dumb autocorrect chatbot" clearly doesn't hold. 

[Natasha Tori Maru]

@Butters if you read over the points carefully you will find your synopsis is a reductive mischaracterisation

[Leo Gura]

8 hours ago, UnbornTao said:

@Hiluy It'd be hilarious if you were a bot.

The movie should be written and made by AI too, for max integrity 

2 hours ago, vinc3nc said:

If it drops to 40k, that's already deep value, yet you still won't buy because you'll be waiting for 30k or even 20k.

But the difference is the stock market cycle. Don't buy at the top of the cycle. If there was no looming bubble, buying Bitcoin would be okay. But buying it today is not worth the risk. I would rather buy gold instead.

36 minutes ago, Butters said:

How can you debate the usefulness of AI when it literally changed warfare? Fully autonomous drones in the Ukraine and Iran wars is now the standard. So"it's just a dumb autocorrect chatbot" clearly doesn't hold. 

That is different from chatbots. Chatbots cannot sustain trillions of investment. Other kinds of AI are useful and worth investing in.

Again, this is not about being anti-AI, it is about anti-hype.

Edited 16 minutes ago by Leo Gura

[Butters]

4 minutes ago, Leo Gura said:

That is different from chatbots. Chatbots cannot sustain trillions of investment. Other kinds of AI are useful and worth investing in.

But the actual serous usage of the OpenAI and Claude technology is through their coding agents, not the chatbots. There are workflows and agents eating up thousands of times the tokens of someone just chatting. That's where the actual heavy usage is at.