271–280 of 387 posts
Earlier quoted context omitted.
Do the activations tell you anything more than what the LLM delivers in plain text? Other than for trivial bugs in the LLM code, I don't think so.
Yes, "making up an answer" will look different from "quoting pretrained knowledge" because eg the model might've decided you were asking a creative writing question.
Earlier quoted context omitted.
Google’s transformation of conventional methods into means of hypercapitalist surveillance is both pervasive and insidious. The “normal definition of that term” hides this.
You don't need "hypercapitalist surveillance" to show someone ads for a PS5 when they search for "buy PS5". If they're doing surveillance they're not doing a good job of it, I make no effort to hide from them and approximately none of their ads are personalized to me. They are instead personalized to the search results instead of what they know from my history. Meta is the one with highly personalized ads.
If simple keyword matching was enough, why would they spend literally billions a year on other tactics?
It's debatable whether Google has truly lost the plot because of the "AI wars", but the moment the statement "Bing returns more sensible results than you" becomes verifiably true, it's... cause for concern? The approach that Google appears to have taken, which is to assume that the top-ranked part of its current search index is a sensible knowledge base, may have been true some years ago, but definitely isn't now: fo…
"Achieving the initial 80 percent is relatively straightforward since it involves approximating a large amount of human data, Marcus said, but the final 20 percent is extremely challenging. In fact, Marcus thinks that last 20 percent might be the hardest thing of all." 100% completely accurate is super-AI-complete. No human can meet that goal either. No, not even you, dear person reading this. You are wrong about som…
Earlier quoted context omitted.
I feel like there's some semantic slippage around the meaning of the word "accuracy" here. I grant you, my print Encyclopedia Britannica is not 100% accurate. But the difference between it and a LLM is not just a matter of degree: there's a "chain of custody" to information that just isn't there with a LLM. Philosophers have a working definition of knowledge as being (at least†) "justified true belief." Even if a LLM…
Encyclopedia Britannica is also wrong in a reproducible and fixable way. And the input queries a finite set. It's output does not change due to random or arbitrary things. It is actually possible to verify. LLMs so far seem to be entirely unverifiable.
"Achieving the initial 80 percent is relatively straightforward since it involves approximating a large amount of human data, Marcus said, but the final 20 percent is extremely challenging. In fact, Marcus thinks that last 20 percent might be the hardest thing of all." 100% completely accurate is super-AI-complete. No human can meet that goal either. No, not even you, dear person reading this. You are wrong about som…
Or to put it another way, I think Google should have a way of saying "yes, we know this result is wrong, but we're leaving it in because it's funny." There is a demand for funny results. Someone asking “how many rocks should I eat” is looking for entertainment, so you might as well give it to them.
These specific results aren't the problem, though. They're illustrations of a larger problem -- if a single satirical article or Reddit comment can fool the model into saying "eating rocks is good for you" or "put glue in your pizza sauce", there are certain to be many more subtle inaccuracies (or deliberate untruths) which their model has picked up from user-generated content which it'll regurgitate given the right prompt.
Earlier quoted context omitted.
that's the part that scares me. I railed on someone's comment the other day about "indexes will come back into fashion" but the more I think about how much garbage has increased in just the past 2 to 3 years, I think I was wrong. Indexes and forums may be the only way to have a sane net where you can find things. Perhaps communities linking together in a ring like format, a "web ring" of sorts.
What I've been wanting to see for a while now is a social-network based search engine: * No pages are indexed automatically. The only indexed pages are pages that users say are worth indexing. Probably have a browser add-on for a one button click that people can use. * You can friend/follow others * Your search results are a combination of your own indexed pages and the pages indexed by people in your network.
This approach to remove bad search suggestions manually reminded of a different approach Google once took, where they weren’t satisfied with manually tweaking search results but rather wanted to tweak the algorithm that produces these results when there were bad results. 'Around 2002, a team was testing a subset of search limited to products, called Froogle. But one problem was so glaring that the team wasn't comfort…
What stops the vendor, or other vendors, from creating more gnomes with sneakers. Easy money from customer with billions of dollars to spend on payola, fines, legal settlements, etc.
Maybe they made the vendor sign an NDA.
Earlier quoted context omitted.
> it will give you a logical stepwise progression from your question to its answer. No, it will generate a new hallucination that might be a logical stepwise progression from the question you asked to the answer you gave, but it is not due to any actual internal reasoning being done by the LLM
So what? You have no way to know for sure if the human you ask the same question, does either. The question that started this thread was related to verifiability. And i still think it is a spurious complaint, given that we have exactly the same limitations when dealing with any human agent.
The human might lie, but they generally don't.
An LLM is always confabulating when it explains how it reached a conclusion, because that information was discarded as soon as it picked a word.
The limitations are not in the same ballpark.