ByteJunk @ ByteJunk @lemmy.world

Posts

2

Comments

1,330

Joined

2 yr. ago

But then how can you tell that it's not an actual conscious being?

This is the whole plot of so many sci-fi novels.

I'll bite.

How would you distinguish a sufficiently advanced word calculator from an actual intelligent, conscious agent?

Let me grab all your downvotes by making counterpoints to this article.

I'm not saying that it's not right to bash the fake hype that the likes of altman and alienberg are making with their outlandish claims that AGI is around the corner and that LLM are its precursor. I think that's 100% spot on.

But the news article is trying to offer an opinion as if it's a scientific truth, and this is not acceptable either.

The basis for the article is the supposed "cutting-edge research" that shows language is not the same as intelligence. The problem is that they're referring to a publication from last year that is basically an op-ed, where the authors go over existing literature and theories to cement their view that language is a communication tool and not the foundation of thought.

The original authors do acknowledge that the growth in human intelligence is tightly related to language, yet assert that language is overall a manifestation of intelligence and not a prerequisite.

The nature of human intelligence is a much debated topic, and this doesn't particularly add to the existing theories.

Even if we accept the authors' views, then one might question if LLMs are the path to AGI. Obviously many lead researchers in AI have the same question - most notably, Prof LeCun is leaving Meta precisely because he has the same doubts and wants to progress his research through a different path.

But the problem is that the Verge article then goes on to conclude the following:

an AI system might remix and recycle our knowledge in interesting ways. But that’s all it will be able to do. It will be forever trapped in the vocabulary we’ve encoded in our data and trained it upon — a dead-metaphor machine. And actual humans — thinking and reasoning and using language to communicate our thoughts to one another — will remain at the forefront of transforming our understanding of the world.

This conclusion is a non sequitur. It generalizes a specific point about the capacity of LLMs to evolve into true AGI or not, into an "AI dumb" catchall that ignores even the most basic evidence that they themselves give - like being able to "solve" go, or play chess in a way that no human can even comprehend - and, to top it off, conclude that "it will never be able to" in the future.

Looking back at the last 2 years, I don't think anyone can predict what AI research breakthroughs might happen in the next 2, let alone "forever".

Very. But the main issue is that there's no wind or water around, even gravity is wimpy. This compounds the problem.

So, on one hand, there's no erosion to smooth the sand particles, so it's not really sand, it's more like very small shards of very sharp glass, that stay that way forever.

Since there's no wind or water or even proper gravity, once it sticks to you, it's not letting go. Even minimal electrostatic forces make you a magnet for these sharp shards, so now you're coated in a layer of extremely abrasive particles that you're not getting rid of, and eventually it'll get into your suit and kill you.

I really feel the need to share this concern, especially because I'll probably forget it soon, but I see that somehow I missed what you were saying, and if I share this fascinating fact with you now I'm in even deeper trouble, so I'll just take the loss here and apologise...

I heard every single word you said, I swear. My brain just decided that it needed that space to ponder just how fine moon dust is while you were speaking, so your audio was never processed.

But we can try again if you want!

I've worked at large (5k+ workers) companies that were running Windows XP well into the late 2010's, with matching hardware. That was too extreme (goddamn ie6).

But this article makes me sick. If the economy needs people to throw away perfectly usable goods and buy new ones, the problem isn't the people, it's the fucking economy. It's time to take a step back and rethink the system, because it's gonna implode.

Same. I don't think he is my brother.

I'll make a bold prediction that we won't have 25 months this year either. Maybe next century.

I had a lousy impression of the US school system, but I never realized it was this bad...

A more serious answer - it depends greatly on where I'm working and what we're doing.

I've worked in places where we'd receive outsource work. Usually we'd get fairly detailed instructions about what to do and what to avoid, that were discussed between our PMs/architects and the client, including tests for example that were agreed upon. You were supposed to follow those to the letter, but the most important part was that you needed to deliver quickly because the customer wanted to keep costs to a minimum. "Useless questions" (from their perspective) were seriously frowned upon, so if it wasn't specified, the expected approach was to do whatever was quicker.

This occasionally lead to situations where their QC/UATs would identify issues with their business rules, but as long as it was compliant with the requirements we received, it would then come back to be changed (at additional cost, depending on how big the change needed to be).

Once accepted though, job done, grab your next work item and move on. Months later they could run into a situation like the one in the printer and come back asking for a fix, but very likely that would go into the CR bucket and a quote would be provided.

Of course if you're working for a company that actually cares about what they're building, the philosophy is completely different. If I'm working on our products, then I build a good understanding of what I'm working on, and I'm expected to flag any concerns or issues I encounter even before it reaches QC.

That said, I've never heard of a developer ever being criminally charged other than intentional misconduct - like, in the world. Look at the IBM Queensland Health payroll system fiasco, I'm not sure anyone was even fired, let alone prosecuted.

Or even the Boeing 737-MAX crashes - how do you build a system that pitches the nose down repeatedly, without limitations? Those guys who worked on the MCAS software would 100% have considered a scenario where an angle-of-attack sensor would provide bad data, and the consequences of repeated trim, but alas - 2 planes crashed, 350 people died, and what are the consequences? Some payouts...

Look around you, you'll find "unrestricted fields in a public-facing app" (from a practical perspective) everywhere. Shrek's script has what, less than 50k characters? That's nothing, you can fit that in a Facebook post and still have more than enough to write a full movie review.

Where this would likely raise flags is when somebody decided that it needs to be printed, but that could be a different team, maybe outsourced, maybe after the main app was developed, maybe it's just some "plug-and-play" system that also handles bulk printing jobs, who knows.

100% accurate, though vibe coding is optional.

If I have a set of requirements that don't mention any type of restriction, then I won't arbitrarily add one - as far as I know, I could be breaking intended functionality. If I'm invested in this, I'll add it to the list of stuff that needs clarification, otherwise it's gonna ship as specified, and eventually someone's gonna file a change request.

That's great, could you perhaps schedule a follow-up since I have a hard stop in 1 min? Thank you, you're the best, see you everyone!

I'm not doing overtime my man.

The waiters still check, for some reason. I always order a cola here, and still get "is pepsi ok?"

Reuters' fact checked it and called it false.

https://www.reuters.com/article/world/-anti-marijuana-advert-that-warns-about-the-devils-lettuce-was-not-created-b-idUSKBN28P2DU/

I like this. I mean, the fact that the rules assist the narrative, but they're not the narrative themselves.

For the desintegrate situation, I'd love for the GM to go something like:

"As you speak the final words of your incantation, Wizard, a thin green ray begins to form on your fingertip. The villain merely smirks, clicking his fingers. A wave of crimson energy smothers your hand, and your spell snuffs out like a candle. He brushes a piece of dust from his shoulder. 'Impatience. Such a childish trait. As I was saying...'"

The GM wouldn't even explain what happened, just continue his narrative, and at some point the party would find that one of the nearby minions in hiding had a counterspell ready, for example.

That sounds unreasonable. If you're stepping up to known thugs to threaten them with violence, violence is expected by the party. There should be no surprise here.

Says the guy holding the fucking jerrycan.

There's a massive cloud of smoke coming from the woods, we caught a guy coming out from there, we're asking him to show us if that gasoline jerrycan-shaped object in the plastic bag that he's holding is in fact a gas jerrycan, but he's still refusing.

And you come out from the toilets "but how do you know that there's a fire"?

It's like they could create a socialist party, so that America actually has at least one left-leaning party.

No, the mainline Democratic party is nowhere near being left wing.