Because its capacities are focused on exactly the right place to feel magical. Which isn’t to say that there isn’t real utility, but language (written, and even moreso spoken) has an enormous emotional resonance for humans, so this is laser-targeted in an area where every advance is going to “feel magical” whether or not it moves the needle much on practical utility; it’s not unlike the effect of TV news making you feel informed, even though time spent watching it negatively correlates with understanding of current events.
Kind of this. That was one of the themes of the movie Westworld where the AI in the robots seemed magical until it was creepy.
I worry about the 'cheery intern' response becoming something of a punch line.
"Hey siri, launch the nuclear missiles to end the world."
"That's a GREAT idea, I'll get right on that! Is there anything else I can help you with?"
Kind of punch lines.
Will be interesting to see where that goes once you've got a good handle on capturing the part of speech that isn't "words" so much as it is inflection and delivery. I am interested in a speech model that can differentiate between "I would hate to have something happen to this store." as a compliment coming from a customer and as a threat coming from an extortionist.
This is basically just the ship computer from Hitchhikers Guide to the Galaxy.
"Guys, I am just pleased as punch to inform you that there are two thermo-nuclear missiles headed this way... if you don't mind, I'm gonna go ahead and take evasive action."
ChatGPT is now powered by Genuine People Personality™ and OpenAI is turning into the Sirius Cybernetics Corporation (who according to the HHGTTG were "a bunch of mindless jerks who were the first against the wall when the revolution came")
I did wonder if there's a less verbose mode. I hope that's not a paywalled feature. Honestly it's possible that they use the friendliness to help buy the LLM time before it has to substantively respond to the user.
Yeah people around me here in Central Europe are very sick of that already. Everybody is complaining about it and the first thing they say to the bot is to cut it out, stop apologizing, stop explaining and get to the point as concisely as possible. Me too.
Or perhaps the news media has been increasingly effective at convincing us the world is terrible. Perceptions have become measurably detached from reality:
Or perhaps the reality on the ground for the working and middle class masses is not the same reality experienced by the elites, and the upper middle class with $150K+ salaries, or measured by stock market growth, and such...
As John Stewart says in https://www.youtube.com/watch?v=20TAkcy3aBY - "How about I hold the fort on making peanut butter sandwiches, because that is something I can do. How about we let AI solve this world climate problem".
Yet to see a true "killer" feature of AI, that isn't doing a job badly which humans can already do badly.
>the point of all of this is: this is alpha 0.45 made to get the money needed to build AGI whatever that is
Maybe they publicly made it available at alpha 0.7 and now it's more like 0.9 RC instead, with not much room to go except through marginal improvements for an ever increasing training budget making them less and less worthy?
And that's before 90% of the internet becomes LLM ouput, poisoning any further corpus for training and getting into LSD-grade hallucinations mode...
It’s not an either-or: the stuff feels magical because it both represents dramatic revelation of capability and because it is heavily optimized to make humans engage in magical thinking.
These things are amazing compared to old-school NLP: the step-change in capability is real.
But we should also keep our wits about us, they are well-Des robed by current or conjectural mathematics, they fail at things dolphins can do, it’s not some AI god and it’s not self-improving.
Let’s have balance on both the magic of the experience and getting past the tech demo stage: every magic trick has a pledge, but I think we’re still working on the prestige.
Yes, the announcement explicitly states that much of the effort for this release was focused on things that make it feel magical (response times, multiple domains, etc.), not on moving the needle on quantifiable practical performance. For future releases, the clever folks at OpenAI are surely focused on improving performance on challenging tasks that practical utility -- while maintaining the "magical feeling."
> explicitly states that much of the effort for this release was focused on things that make it feel magical (response times, multiple domains, etc.), not on moving the needle on quantifiable practical performance.
Hmm, did you mean implicitly? I've yet to see where they say anything to the likes of not "moving the needle on quantifiable practical performance."
Pretty interesting how it turns out that --- contrary to science fiction movies --- talking naturally and modelling language is much easier and was achieved much sooner than solving complex problems or whatever it is that robots in science fiction movies do.
I didn't use it as a textual interface, but as a relational/nondirectional system, trying to ask it to inverse recursive relationships (first/follow sets for BNF grammars). The fact that it could manage to give partially correct answers on such an abstract problem was "coldly" surprising.
You really think OpenAI has researchers figuring out how to drive emergent capabilities based on what markets well?
Edit: Apparently not based on your clarification, instead the researchers don't know any better than to march into a local maxima because they're only human and seek to replicate themselves. I assumed too much good faith.
I don’t think the intent matters, the effect of its capacities being centered where they are is that they trigger certain human biases.
(Arguably, it is the other way around: they aren’t focused on appealing to those biases, but driven by them, in the that the perception of language modeling as a road to real general reasoning is a manifestation of the same bias which makes language capacity be perceived as magical.)
Intent matters when you're being as dismissive as you were.
Not to mention your comment doesn't track at all with the most basic findings they've shared: that adding new modalities increases performance across the board.
They shared that with GPT-4 vs GPT-4V, and the fact this is a faster model than GPT-4V while rivaling it's performance seems like further confirmation of the fact.
-
It seems like you're assigning emotional biases of your own to pretty straightforward science.
> Intent matters when you're being as dismissive as you were.
The GP comment we're all replying to outlines a non-exhaustive list of very good reasons to be highly dismissive of LLM. (No I'm not calling it AI, it is not fucking AI)
It is utterly laughable and infuriating that you're assigning legitimate skepticism about this technology as a an emotional bias. Fucking ridiculous. We're now almost a full year into the full bore open hype cycle of LLM. Where's all the LLM products? Where's the market penetration? Business can't use it because it has a nasty tendency to make shit up when it's talking. Various companies and individuals are being sued because generative art is stealing from artists. Code generators are hitting walls of usability so steep, you're better off just writing the damn code yourself.
We keep hearing this "it will do!" "it's coming!" "just think of what it can do soon!" on and on and on, and it just keeps... not doing any of it. It keeps hallucinating untrue facts, it keeps getting basics of it's tasks wrong, for fucks sake AI Dungeon can't even remember if I'm in Hyrule or Night City. Progress seems fewer and farther between, with most advances being just getting the compute cost down, because NO business currently using LLM extensively could be profitable without generous donation of compute from large corporations like Microsoft.
I didn't see any good reasons to be dismissive of LLMs, I saw a weak attempt at implying we're at a local maxima because scientists don't know better than to chase after what seems magical or special to them due to their bias as humans.
It's not an especially insightful or sound argument imo, and neither are random complaints about capabilities of systems millions of people use daily despite your own claims.
And for the record:
> because NO business currently using LLM extensively could be profitable without generous donation of compute from large corporations like Microsoft
OpenAI isn't the only provider of LLMs. Plenty of businesses are using providers that provide their services profitably, and I'm not convinced that OpenAI themselves are subsidising these capabilities as strongly as they once did.
All that spilled ink don’t change the fact that I use it every day and it makes everything faster and easier and more enjoyable. I’m absolutely chuffed to put my phone on a stand so GPT4o can see the page I’m writing on and chat with me about my notes or the book I’m reading and the occasional doodle. One of the first things I’ll try out is to see if it can give feedback and tips on sketching, since it can generate images with a lot better control of the subject it might even be able to demonstrate various techniques I could employ!
As it turns out, people will gleefully welcome Big Brother with open arms as long as it speaks with a vaguely nice tone and compliments the stuff it can see.
A year is an eternity in tech and you bloody well know it. A year into an $80 billion dollar valued company's prime hype cycle, and we have... chatbots, but fancier? This is completely detached from sanity.
I mean when you’re making a point about how your views should not be taken as emotional bias, it pays to not be overly emotional.
The fact that you don’t see utility doesn’t mean it is not helpful to others.
A recent example, I used Grok to write me an outline of a paper regarding military and civilian emergency response as part of a refresher class.
To test it out we fed it scenario questions and saw how it compared to our classmates responses. All people with decades of emergency management experience.
The results were shocking. It was able to successfully navigate a large scale emergency management problem and get it (mostly) right.
I could see a not so distant future where we become QA checkers for our AI overlords.
>they aren’t focused on appealing to those biases, but driven by them, in the that the perception of language modeling...
So yes in effect that is their point, except they find the scientists are actually compelled by what markets well, rather than intentionally going after what markets well... which is frankly even less flattering. Like researchers who enabled this just didn't know better than to be seduced by some underlying human bias into a local maxima.
I think that's still just an explanation of biases that go into development direction. I don't view that as a criticism but an observation. We use LLMs in our products, and I use them daily and I'm not sure how that's that negative.
We all have biases in how we determine intelligence, capability, and accuracy. Our biases color our trust and ability to retain information. There's a wealth of research around it. We're all susceptible to these biases. Being a researcher doesn't exclude you from the experience of being human.
Our biases influence how we measure things, which in turn influences how things behave. I don't see why you're so upset by that pretty obvious observation.
The full comment is right there, we don't need to seance what the rest of it was or remix it.
> Arguably, it is the other way around: they aren’t focused on appealing to those biases, but driven by them, in the that the perception of language modeling as a road to real general reasoning is a manifestation of the same bias which makes language capacity be perceived as magical
There's no charitable reading of this that doesn't give the researcher's way too little credit given the results of the direction they've chosen.
This has nothing to do with biases and emotion, I'm not sure why some people need it to be: modalities have progressed in order of how easy they are to wrangle data on: text => image => audio => video.
We've seen that training on more tokens improves performance, we've seen that training on new modalities improves performance on the prior modalities.
It's so needlessly dismissive to act like you have this mystical insight into a grave error these people are making, and they're just seeking to replicate human language out of folly, when you're ignoring table stakes for their underlying works to start with.
Note that there is only one thing about the research that I have said is arguably influenced by the bias in question, “the perception of language modeling as a road to real general reasoning”. Not the order of progression through modalities. Not the perception that language, image, audio, or video are useful domains.
Because its capacities are focused on exactly the right place to feel magical. Which isn’t to say that there isn’t real utility, but language (written, and even moreso spoken) has an enormous emotional resonance for humans, so this is laser-targeted in an area where every advance is going to “feel magical” whether or not it moves the needle much on practical utility; it’s not unlike the effect of TV news making you feel informed, even though time spent watching it negatively correlates with understanding of current events.