drob518

9 ms·

211.
by drob518 4mo ago
I think that’s part of the commenter’s point.
212.
by drob518 4mo ago
Which are what, exactly?
213.
by drob518 4mo ago
Why can’t AIs generate for the “existing, mature ones?” Like the other commenter said, I’m not sure I get the “this is totally for AI” marketing. Why can’t AI use the existing ones and why can’t humans use this?
214.
by drob518 4mo ago
I went through a similar phase decades ago with Common Lisp. It takes a week or two. Now, it’s quite a natural syntax and I see the parens as a huge benefit. I like Clojure syntax even more than CL and Scheme because of the map and vector l
215.
by drob518 4mo ago
Like what?
216.
by drob518 4mo ago
To be sure, social media is a problem for our children. No doubt about it. But the solution is to take screens from the children, not set up the foundation of a police state.
217.
by drob518 4mo ago
In fact, that’s the much bigger threat.
218.
by drob518 4mo ago
Okay, but that still has a limit, right? Do training costs have a limit? Everyone is in the frothy stage of this technology wave and they continue to buy more, but training the next model requires exponential increases in model sizes to get
219.
by drob518 4mo ago
I believe that OpenAI is in trouble. Too many high level defectors. Altman is clearly in hype mode. Is Amodei? Yes, I’m pretty sure he is at some level. He might also believe it partially (it’s not totally binary).
220.
by drob518 4mo ago
It’s all about timing. This is tech bubble 2.0, Dotcom Boogaloo. If you’re able to flip it quickly, you’ll have generational wealth. If not, you could be holding a lot of worthless paper.
221.
by drob518 4mo ago
Inference is the phase where they make money. But the question is whether they can be profitable overall as training continues to balloon.
222.
by drob518 4mo ago
Even if they aren’t pushing it yet, as soon as everyone is identified and characterized, the data exists and can be used for anything.
223.
by drob518 4mo ago
It’s not for the kids. That’s just the excuse. In order to validate kids, they really need to identify everyone. That’s the real play.
224.
by drob518 4mo ago
Well, that was depressing.
225.
by drob518 4mo ago
Perhaps, but as the AI analysis becomes part of the release process (or even the CI process as prices fall), you’d expect those new issues to be caught before release and fixed. We’re seeing them caught post-release for now because the code
226.
by drob518 4mo ago
Yes, of course. I’m definitely anthropomorphizing as a shorthand. I’m the first one to say that these models are just a lot of matrix math.
227.
by drob518 4mo ago
There is going to be a flurry of this sort of stuff as the AIs get smart enough to find them. It will naturally die down as the legitimate ones are fixed. Yes, there will always be some level of this, but I’d expect it to be low and the exp
228.
by drob518 4mo ago
So sad. I first met Om in 2001 or so. I pissed him off because I wouldn’t meet with him to do an interview for our startup. He always loved getting the early scoop and we weren’t ready for any publicity. In later years, we would laugh about
229.
by drob518 4mo ago
Well done. Here’s hoping that a hand-written GPU backend gets developed based on this wonderful proof of concept. There’s no reason to not take advantage of the state of the art hardware when it’s available. And screens are only moving towa
230.
by drob518 4mo ago
I’m sure they will license it. It’s better for them if everyone in the industry can innovate on everything around it. All the process tech companies will make it more cost effective, for instance, which helps IBM as well.
231.
by drob518 4mo ago
Absolutely. And why would you? Companies spend lots of time talking about loyalty and teamwork, but they show their colors when they do these layoffs. Smaller companies, often still run by the founder, can be much better. The only large com
232.
by drob518 4mo ago
Sometimes you do get the CEO gig for being a pioneer, but then the whole organization thwarts whatever you want to do by repeatedly saying “we don’t do it that way here” and dragging their feet until you get fired.
233.
by drob518 4mo ago
And the board is pulling down $300k per year or more for sitting in maybe 11 meetings per year and participating in a “comp committee” where they just review data and recommendations from comp consultants and agree to whatever the consultan
234.
by drob518 4mo ago
Let’s be honest about how the incentives work at large companies. The CEO probably has a $10m/yr comp package. The EVPs under him are $3m-$5m each. Nobody is really interested in making the company wildly successful, because that would
235.
by drob518 4mo ago
There is a huge cost for this either way (severance packages, yes, but also lost productivity, reduced team coherence, etc), but that unfortunately doesn’t necessarily translate to a political cost for the managers involved in pushing the d
236.
by drob518 4mo ago
Careful. I don’t have anti-Musk bias, but it can both be true that SpaceX will be quite successful in the long run and the stock is still overpriced in the short run.
237.
by drob518 4mo ago
Look, if you’re committed to Blub, don’t let me talk you out of it.
238.
by drob518 4mo ago
My editor balances my parentheses (Emacs Paredit). I rarely think about them. I just think structurally and the editor manages the details.
239.
by drob518 4mo ago
Oh, it’s “for agents,” you say. Well, I’m sure that makes all the difference.
240.
by drob518 4mo ago
Yep, I see it.

More ›