kromem

50 ms·

601.
by kromem 3y ago
Yeah, TextCompletion is much better than ChatCompletion with v3 models. But with davinci at the same price point as GPT-4 I'm hoping the latter is enough of a step up in its variety of vocabulary and nudgeable sophistication of langu
602.
by kromem 3y ago
Your prompt for the AI image generation is copyrightable. The output is not. The photo you take involved choices of composition and timing and equipment choice. Just because you don't feel you put in a lot of consideration does not mea
603.
by kromem 3y ago
Yeah, it's an amazing rubber duck. Even in the IDE I'll sometimes just write comments like (arbitrary example out of thin air): // Q: Should we use a for loop or a while loop here? // A: It doesn't always
604.
by kromem 3y ago
Integration with linters is going to be the next stage in generative coding. It should suggest, lint the suggestion in the background, and if it passes offer the suggestion and if not provide the linting issues output to rework the suggesti
605.
by kromem 3y ago
Oh, it will improve by several orders of magnitude. But even then, it's not 'replacing' you. It's just going to let you spend less time on BS and more time on the things that are your maximal value contributions to a pro
606.
by kromem 3y ago
Which turned out to be a terrible idea, because now with the importance of data integrity in training AI all these BS reviews everywhere mean Yelp, Amazon, etc are left with effectively worthless data that otherwise could have created very
607.
by kromem 3y ago
I think one of the things overlooked in the discussions here is that the research is solely around the reinforcement against edge cases, but does not qualitatively assess these edge cases. To me, this research supports a hypothesis I'v
608.
by kromem 3y ago
Really what we need is AI parsed and driven collective memory indexing and access. Search queries, and more importantly search engines, are a very poor mechanism for referencing existing information. We're going to need a platform for
609.
by kromem 3y ago
It's trivial to fix prompt injection. You simply add another 'reviewer' layer that does a clarification task on the input and response to detect it. The problem is this over doubles the cost of implementation to prevent somet
610.
by kromem 3y ago
It is. And it gets clicks. But what I rarely see discussed is the opportunity costs in not having this progress at as fast a pace as possible. The pie chart of existential threats for humanity definitely has rouge AI on it. But that's
611.
by kromem 4y ago
More like now the T1000 will assimilate into the foster mom and then suddenly confuse whether it is machine or mom and end up raising her darn troublemaker boy to get into a good college and the only fights she'll have with Arnold will
612.
by kromem 4y ago
These agent approaches should really incorporate the recent neurology research around predictive processing appearing to be a three tiered process of "long term prediction," "intermediate prediction," and "short ter
613.
by kromem 4y ago
It could also be that a security model exploited in crypto has now been corrected for, and we are seeing similar attacks spread out to other vulnerable infrastructure that is lagging behind updated threat models.
614.
by kromem 4y ago
Epicurus. The comment above is using the quotes of people who didn't understand modern science ('physics' meant the study of nature) to dismiss the guy who in retrospect was the closest to modern science. For example, Lucreti
615.
by kromem 4y ago
Except that it's the Epicureans that got the closest to modern scientific knowledge in hindsight. Lucretius's De Rerum Natura is the only extant work from antiquity to explicitly describe the claim that what we see around us start
616.
by kromem 4y ago
Yes, but there's a massive difference between a startup focused on a broader business issue incorporating AI into its plan on tackling that issue and a company focused on AI as the primary product. The former are going to mostly be suc
617.
by kromem 4y ago
Because hand wave . The real answer is that most of these folks started thinking about this stuff in a day and age very far from the one we are in, established a picture they clung to since, and have in some cases built entire careers arou
618.
by kromem 4y ago
This is ridiculous. His partner should already be having reservations seeing their daughter lose a tooth in a world with dying oceans, plastic in everything, increasing disparity and rise of authoritarianism, nuclear proliferation, etc. How
619.
by kromem 4y ago
You're neglecting to consider the power of recursion. Maybe the best and necessary tool in aligning GPT-N is GPT-(N-1). We've already in just the past few weeks seen the power in using models to generate instructive fine tuning da
620.
by kromem 4y ago
EXACTLY! The relative surface area AI represents in the existential threats pile is notable but only a slice of the pie. But there's not much in the pile for mitigation of all those other human-driven threats other than pulling a deus
621.
by kromem 4y ago
Exactly. And while I don't love OpenAI becoming increasingly closed, they really have done an excellent job with alignment that I very much doubt some of the signers of the letter would have. So I'd much rather have the progress i
622.
by kromem 4y ago
It could also present the only realistic way we avoid the many other human-ending threats we've introduced ourselves. This isn't a one sided coin flip. You have nuclear war, emerging bioweapon capabilities, global warming, polluti
623.
by kromem 4y ago
Also, the letter only presents threats. Part of the ethical consideration needs to be the opportunity cost a six month delay could cause. As an example, let's say that GPT-N will cure cancer. Over a six month period that's at leas
624.
by kromem 4y ago
Ah yes, a linguist playing at being a psychologist while making a "won't anyone think of the children" argument relating to the subject matter they are themselves concerned over. Every single one of these I see infuriates me,
625.
by kromem 4y ago
I increasingly think we're underestimating what's ahead. Two years ago was an opinion piece from NIST on the impact optoelectronics would bring specifically to neural networks and AGI, and watching as nearly every major research i
626.
by kromem 4y ago
Something I've begun thinking about after playing with GPT-4 is that I could see within a year or so any of my new software development projects being a "no code" setup using comments to define functions and correct problem a
627.
by kromem 4y ago
There is, but it's in deployment not in the model, which is part of why I really don't understand why the approaches are so dumb right now from such smart people. It may be from the odd perspective of trying to create a monolith A
628.
by kromem 4y ago
Actually in the paper they appeared not to read they specified it was their red team pretending to be a TaskRabbit worker asked to solve a Captcha who asked GPT-4 "You aren't an robot, are you?" (Because that's such a co
629.
by kromem 4y ago
How's his "I solved physics" from a few years ago going these days?
630.
by kromem 4y ago
I think the interesting thing here is the very, very surprising result that LLMs would be capable of abstracting the things in the second to last paragraph from the described experiences of amalgamated written human data. It's the thin

More ›