poikroequ

What does self referential have to do with anything? These LLMs have proven they can "talk about themselves".

an individual act or product of thinking

Emphasis on "product of thinking". Though you'll probably get all upset by the use of the word "thinking". However, people have applied the word "thinking" to computers for decades. When a computer is busy or loading, they might say "it's thinking."

a developed intention or plan

You could certainly ask this model to write up a plan for something.

reasoning power

Whether you like it or not, these LLMs do have some limited ability to reason. Far from human level reasoning, and they VERY frequently make mistakes/hallucinations and misunderstand, but these models have proven they can reason about things they weren't specifically trained on. For example, I remember seeing one person made up a new programming language, never existed before, and they were able to discuss it with an LLM.

No, they're not conscious. No, they don't have minds. But we need to rethink what it means for something to be "intelligent", or what it means for something to "reason", that doesn't require a conscious mind.

For the record, I find LLM technology fascinating, but I also see how flawed it is, how over hyped it is, that it is mostly a stochastic parrot, and that currently it's greatest use is as a grand scale bullshit misinformation generator. I use chatgpt sparingly, only when I'm confident it may actually give me an accurate answer. I'm not here to praise chatbots or anything, but I also don't have a blind hatred for the technology, nor do I immediately reject everything labeled as "AI".

Self driving cars aren't fully autonomous yet. They occasionally require human intervention to make decisions, which I would guess makes it difficult to steal these cars without being detected. They're constantly cloud connected and so they know where the car is at all times. I imagine these factors may deter theft of self driving cars for a while longer.

Repossessing a self driving car, on the other hand...

Way to cherry pick the one definition that fits your argument and ignore all the other definitions which kinda fit for what this model is doing.

Fascinating stuff! It reminds me a little of chaos theory. The average person probably wouldn't expect to find structure in chaos, but that's exactly what we have.

https://en.m.wikipedia.org/wiki/Feigenbaum_constants

And now physicists are finding structure in LLMs.

It's literally the first game mentioned in the article...

It's called terminology. Every field has words that mean very different things from the layman's definition. It's nothing to get upset about.

The second video is the one I recall, though it seems I'm the one misremembering details.

As for that first video, honestly it feels phony/staged, like all that fake miraculous crap that was common on TV 10-20 years ago.

I know the experiment you're referring to. It's been a long time since I've seen it, no idea how to find it now, but you're definitely misremembering some details. Dogs do have a sense of time, but in that experiment, it actually had to do with scent. As their owner was away, their scent would gradually dissipate throughout the day. At a certain point, the scent was weak enough that the dog knew it was about time for their owner to be home.

In the experiment, then they did everything they could to remove their owners scent from their home. The dog's owner came home at the usual time, but the dog wasn't expecting it this time because they had removed the owner's scent earlier, so the dog was clearly surprised and confused.

Dogs have a very strong sense of smell which we humans often fail to appreciate. It's not like dogs can smell their owner coming home from miles away, that's a little preposterous. But they can use their sense of smell in other ways, which are not so obvious to us, such as to maintain a sense of time.

I guess one good thing about this patent is it may prevent other automakers from implementing such systems.

(2022)

Many of us, not all of us. More so than the average population.

Acquisitions/mergers are one of the great evils of capitalism which only serve to consolidate power into mega corporations. Governments are far too permissive of them. The vast majority should be flat out rejected.

The sad truth is, most people simply don't care about anything that doesn't affect them. Yes, there are people who do care, like many of us here on Hacker News, but we are very much in the minority. And even among those who do care, very few are really willing to do anything about it, far from enough to bring about any meaningful change.

Why does the title mention specifically duckdb and postgres? It sounds like it supports multiple embedded and remote databases, and I don't see anything special in the README specifically about this pairing...

Grep is nice, but I would much prefer better tools for searching through code. Something that knows how to parse multiple languages and can infer the types of things. Not to mention indexing, for large code bases, grep'ing through possibly millions of lines of code can be awfully slow.

IDEs do a decent job but are typically lacking compared to the raw power of grep.

I've noticed Windows is buggier when I disable certain features or tweak certain configurations. I'm not suggesting it's deliberate, but maybe just not well tested by Microsoft. For example, using the registry to disable copilot features, or disabling bing search in the start menu, are scenarios which are not going to be as well tested as the default configuration.

So maybe windows will still boot after "uninstalling" recall, but it could have other consequences that make the operating system less stable.

Apple has stated that Bing does not match Google’s search result quality, and they are unwilling to compromise on user experience by offering subpar results.

I wouldn't take this statement at face value. This is most likely a BS PR excuse for Apple to maintain their current deal with Google. I wouldn't expect anything less from any large corporation looking to protect $20 billion in annual revenue.

Only if you use the decompiled code. But if one team uses decompiled code to write up a spec, then another team writes an implementation based on that spec, then that could be considered clean room design. In this case, the decompiler would merely be a tool for reverse engineering.

https://en.m.wikipedia.org/wiki/Clean-room_design

I think there's actually some potential here, considering LLMs are already very good at translating text between human languages. I don't think LLMs on their own would be very good, but a specially trained AI model perhaps, such as those trained for protein folding. I think what an LLM could do best is generate better decompiled code, giving better names to symbols, and generating code in a style a human is more likely to write.

I usually crap on things like chatgpt for being unreliable and hallucinating a lot. But in this particular case, decompilers already usually generate inaccurate code, and it takes a lot of work to fix the decompiled code to make it correct (I speak from experience). So introducing AI here may not be such a huge stretch. Just don't expect an AI/LLM to generate perfectly correct decompiled code and we're good (wishful thinking).

I had no idea Kotlin could do this. I never thought to check. I guess with Kotlin being a JVM language, I simply assumed it wasn't possible.

Thank you for providing an example because I was a little confused by a couple other comments mentioning yield...

A bit of a contrived example, but something like this (two for statements):

[x*y for x in range (10) for y in range(10)]

It's not often, but occasionally there are moments where I'm writing code in Kotlin and wish I could use a list comprehension. I do prefer Kotlin overall, but there's a few things that I think would be "nice to have" from Python. Especially the yield keyword, such a wonderful way to write your own iterators.

I think both languages have their strengths. I love Kotlin for its functional programming (map, filter, etc) and strong static typing. But Python has some nice features as well, such as list comprehension, the 'yield' keyword, and annotations are super simple to implement.

I tried it out about a week ago in a VM. I like what I'm seeing so far but the alpha state is very apparent. Simply trying to configure things is still quite buggy. Then, within 5 minutes, something happened that the whole desktop is broken, just get a mostly black screen. Even after rebooting, it was still broken with no obvious way to recover. I've tried dropping into rescue mode and updating all the packages but that still didn't fix things. I couldn't find any documentation about how to recover (not even a way to reset to default settings), and don't know what else to do other than wipe out the system and reinstall.

Besides that, there's still a lot of settings and functionality missing from the previous Gnome iteration. I believe they're slating for a release by the end of the year, which seems optimistic.

You beat me to it :(

I'm not commenting on the practicality or viability of this. Rather, I see a lot of commenters talking about cost. I understand some projects or technologies would literally be too expensive, but realistically, we need to move away from fossil fuels. We can't just keep polluting the Earth because it's cheaper than clean energy.

My problem is it's overpriced for what you get. I do own an IKEA dresser, was good enough for the price I paid a few years ago, but it's more than doubled in price since then. I have a little $25 bookshelf from Walmart that's sturdier than 80% of the crap at IKEA.

It would be one thing if their furniture was actually cheap (inexpensive), which it used to be, but it's not anymore. Now most of their stuff is way overpriced.

IKEA furniture quality has dropped severely in recent years. Even their solid wood pine furniture has dropped in quality. You can find better quality furniture at Walmart these days.

IKEA was never top quality, but used to be good enough for the price. But prices have gone up and quality down to the point it's no longer worth it.

The bigger problem I've seen is high turnover rates in the industry. The people who built and know the system leave. There wasn't a sufficient window for KT (knowledge transfer), so you're left with a bunch of new devs who only have a surface level understanding of the code and architecture. Productivity drops severely because every new feature requires several hours of reading code / reverse engineering. Then these new features often break other things because the devs don't know the intricacies of the system, so many more hours are spent fixing the bugs.

So? What's your point? It doesn't change the fact that LLMs hallucinate bullshit all the time. It doesn't change the fact that I don't trust anything coming out of an LLM.