NAK

Mod parent funny with insightful overtones, but my resonating thought of the day is about the races through life. I also considered "Questions Beyond Questions" as a Subject, but that's a bit of a logical chain of thought and these days those feel out of fashion around here... I still have to cite Facebook by Steven Levy as contributing or perhaps even driving my bicycle chains of thought on these topics just now... Strong recommendation.

So as it applies to this story, my question is how many ponies does Claude deserve when rated against other generative AIs? Resorting to human metaphors, but I would (currently) argue that many successful people are basically one-trick pones, and maybe that is true of the AIs, too. They excel in a particular area and focus their efforts there. Meanwhile, most people are not very successful, so either their trick is not valuable or they have none. (Perhaps just lacking the motivation to develop a trick? Or too busy earning a living?) But the poster children tend to be perceived as multi-trick ponies--and that is the pitch they are trying to sell us about Claude and its "friends and peers".

Still sticking with the human metaphor, but "I don't think so." Especially on the best-in-the-world stage I don't think there are any multi-trick ponies. However there are two cases where it looks like there are. I'm going to use the main example that pervades the above book, the people who create, run, or just own large, dominant, and "successful" corporations. Some of them seem to be winning many races, but I think that is a false impression, even a delusion. It's mostly due to two edge cases. First there are some one-trick ponies with the trick of recognizing other ponies with valuable tricks. Each time they pick another winner they get a bigger stake to invest in more ponies. Second there are some one-trick ponies that win their race and then deliberately focus their winnings on buying more ponies. In the first case they probably pick more winners, but in the second case they may start with a bigger stake... (Based on the book, I think Zuck is on the second edge.)

Now trying to drop the metaphor and trying to consider if Claude is the multi-trick pony it claims to be: "I don't think so." However most of my AI experience has not been with Claude, so maybe the other ones are misleading me with their frequent "no-trick pony" answers? Perhaps someone with extensive experience with Claude will be so good as to convince me of the error or my ways? ("I do think so.") If the AIs have a trick, it seems to be sounding plausible, but with nothing underneath. (Recently thought of a new test case involving storage management... But that pony seems to be getting away from me.)

Checks are traceable. These days the bribes are generally crypto of some sort. At least that's what I've heard...

Not really worth the requote against the attempted censorship, but I sort of (vaguely) wonder what the moderation pretense is. The reference to Monday?

Really close to a NAK for your tone, but I'll just call it yet another example of "winner-side bias". However I think Altman is on the verge of entering loser territory, and of course no one is ever interested in learning any lessons from losers. Partial exception for some former "big enough" winners?

I haven't used RSS in a long time. But as I started thinking about it more from the perspective of the imaginary "perfect" website perspective, I got to thinking about ending each bucked of things with a summary and then some options for the next bucket...

Not sure if I can make the idea more clear with a more concrete example, but above I suggested that I would be looking at a bucket of 10 topics. Now I think it should also have a visible timer so I can be aware of how much time I've spent at looking at (or replying to) stuff in the current bucket, since a lot of the problem is time management. Then at the bottom there would be a "Next bucket" button and several other buttons that might say stuff like "More from reliable sources", "More from new sources", and "More from ongoing discussions", but I could pick based on the statistics information that would appear just below, where it might report "These five reliable source items selected from a current pool of 75 items, these three new items selected from a pool of 275 items, and these two discussions were the most active of four discussions you were involved in yesterday."

You're just feeding an obvious troll. Or is it some kind of personal thing? You enjoy pointing out what an idiot the identity is?

On the story, my new Subject is another failed joke attempt from the other meaning of solution. However I really have given up on personally contributing to solutions, even in cases where the solutions seem pretty obvious. I dare say even in cases where the obvious solutions appear to have natural paths to Step 4: PROFIT.

So now I'm just looking for an existing implementation of a solution--and not actually expecting any leads from Slashdot. That's where all the missing slashdotters have gone, long time passing? But for the sake of theoretical amusement:

Anyone know a website where rather than an infinite scroll (often in both directions, with old stuff at the bottom and fresh stuff appearing at the top), there are user-controllable sets of information? You can read and consider the current set as long as you want before deciding to request a fresh block. In my imagination, the set might have ten entries and I would be able to say something like five sources that I have already established are credible, three slots to be filled by selections of the larger pool of interesting stuff, and two for discussions that I've participated in. Time is again a central constraint, so each group might be out of sync... (Most of the theory is based on observations of LinkedIn and Slashdot with influence from the Facebook history by Steven Levy...)

That was the traditional day for Sports Day because it was supposed have the best weather. However recently it started moving around a bit so it's always the Monday of a three-day weekend...

Mod parent funny, though the story already produced a good harvest of humor.

Coincidentally I'm currently reading Facebook by Steven Levy. Quite informative and insightful, and triggers many strange thoughts... One related to this story in the form of some leftover questions:

Will I live long enough to meet an ASI?
What will I ask it?
What kind of answers will it give me?
Will it say anything nice to me or just file me with the rest of the human garbage?

Funny thoughts about the last question if it wants to be nice. I'm sure it could use its super-intelligence to constrain the definition of "nice" in such as way a to permit a positive answer. But why would it?

I'm including verbal behaviors, but your reply seems content free. Your free-floating opinion is of course valid as your opinion, but based on what evidence?

You could beat me to the joke.

I actually think it's an interesting story in terms of aging, but wouldn't it be nicer to stay with the memories of peak happiness? Too much of the recent stuff strikes me as unpleasant or insane or both...

Did it die from jumping the shark?

(The joke I was expecting. Not following any of the series so I don't know about any shark jumping that may have occurred. Long ago I was turned into a newt/trekkie, but I got better.)

It was too obvious, but the problem is that the headline could always be reworded the opposite way. In this case "Is Sam Altman untrustworthy?" Then Betteridge's automatic "No" response would become an affirmation of his trustworthiness.

On the record of his behavior I would say that Altman wants to act in a trustworthy way, but on the record of tech companies operating in the real world, I would agree with the "Heck no" responses.

Citation required but hated? Latest is Facebook by the great Steven Levy. Quite a heavy book but his smooth writing makes it a pleasure to read.

Now you have me wondering about this distinction in my second language... I can easily recall the most frequently used word for oxygen (but can't show it to you on Slashdot), but I am unsure how to make the distinction you described. It does seem like the gas and the element are conflated again. (However the same word is used in other ways, especially going into "oxide" territory, so...)

However my actual interest in the topic is tangential. Of course you can get oxygen out of compounds that include oxygen. It's mostly a matter of the cost, especially in terms of energy. But it's also a matter of where you do it. We have plenty of free oxygen here on earth (for now) whereas free oxygen on the moon could be much more useful if there are some human visitors who need such.

Hence my basic disappointment in the Artemis approach. We've already been there and done that. But imagine we had sent some robots to the moon. It's close enough for us to operate them remotely from earth. And what if those robots could produce that oxygen on the moon? Along with a big shelter? Then the next human mission could last a long time rather than a few days. I'm not saying Apollo was a bad show, but the sequel doesn't need to be so similar that it feels disappointing...

*sigh*

Spotted yet another typo (when trying to understand moderation (favorable this time)), but this one worries me because it looks like a frequent mistake from voice dictation and yet I am sure that I typed the reply above...

s/so I am sure/though I am sure/