smokel

19 ms·

301.
by smokel 1y ago
Agree. I think it safe to say that we haven't found an optimal solution for the problem discussed in the article. It's not even clear what the reward function should look like, so perhaps the comparison with reinforcement learnin
302.
by smokel 1y ago
Out of curiosity, how many of you are using Kotlin for backend development? It seems that Java has caught up a bit, and the advantages of Kotlin (extension methods, better syntax, better null-checking) might not be enough to justify the ri
303.
by smokel 1y ago
This is related to the exploration–exploitation tradeoff studied in reinforcement learning [1], for which no universally optimal solution is known. This article suggests to explore more, but it might be wise to exploit your knowledge and b
304.
by smokel 1y ago
You seem to be confusing several things. The gold medal IMO is not won by agentic coding software.
305.
by smokel 1y ago
Coding agents are very new. They seem very promising, and a lot of people see some potential value, and are eager to be part of the hype. If you don't like them, simply avoid them and try not to get upset about it. If it's all
306.
by smokel 1y ago
Perhaps we misunderstand each other. I'm saying that the mathematical optimization method differs from the algorithmic paradigm, as explained on this Wikipedia page for example: https://en.wikipedia.org/wiki/Dynam
307.
by smokel 1y ago
Possibly this is related to our individualistic society. The easier it is to project one's own pain onto something, the easier it is to assume that it is a universal problem.
308.
by smokel 1y ago
> Programming used to be (and still is, to a large extent) an activity that can be done with open and free tools. This was largely not the case before 2000, and may not be the case after, say 2030. It may well be that we are living in
309.
by smokel 1y ago
Note that the comment you are replying to is speculating about the (not so distant) future. Be assured that companies will try their best to lock customers in. One option is to add adverts to the generated output, and making the product ch
310.
by smokel 1y ago
Both meanings derive from Bellman. "Reinforcement learning" did not invent a new way, and the field of reinforcement learning predates leetcode by a large margin.
311.
by smokel 1y ago
Also, the term "dynamic programming" has slightly different meanings in the context of leetcode problems and in the context of reinforcement learning. In the first it's for optimizing a relatively small deterministic system,
312.
by smokel 1y ago
> People seem to care much more about... One cannot conclude this from what the market does. Single individuals might want wildly different things than what the combined economy serves them. See also https://en.wikipedia.or
313.
by smokel 1y ago
You may be interested in alternative approaches such as playing chess, running, singing, dancing, or reading literature.
314.
by smokel 1y ago
Oh yes, I fully agree. It will be quite horrible if we don't handle things properly.
315.
by smokel 1y ago
One advantage might be that you would only implement the (small) part of the library that you actually need, potentially avoiding bugs such as Log4Shell. The risk of code duplication is a serious one though, and it would be nice if AI could
316.
by smokel 1y ago
This is only true if you enable extended spell checks, which makes some sense. By default, no form data is sent to Microsoft AFAIK. Note that the same holds for Google Chrome.
317.
by smokel 1y ago
> for some reason It is most likely because we are bad at pattern matching. By default we reward anything we perceive as positive, regardless of who we think is causing it or what the long-term consequences might be. It takes some e
318.
by smokel 1y ago
Why do you think it needs reproducible steps? It is obvious that the bug is still active, so in a way it is reproducible, just not in a systematic way. This happens more often, for example when many services work together in an asynchron
319.
by smokel 1y ago
I notice that some people have become more productive thanks to AI tools, while others are not. My working hypothesis is that people who are fast at scanning lots of text (or code for that matter) have a serious advantage. Being able to di
320.
by smokel 1y ago
This is something where modern chatbots will probably be very helpful. From the top of my head: 1. Validate input, expect all external input (both human and machine generated) to be completely bogus. 2. Fail fast and early. If something s
321.
by smokel 1y ago
Nitpick: not all bugs have to be reproducible to be taken seriously. Defensive programming, and adding extra logging could be a mitigation to avoid future problems, or to help fixing them in the future.
322.
by smokel 1y ago
With ball-point ink, I think (but haven't checked recently) that there is unfortunately no safe option. Some contemporary artists sell digital prints, but if you are selling the real deal, at least ensure that it does not sit in full
323.
by smokel 1y ago
I did similar testing with ball-point pens. Eventually I ran into the ISO 12757-2 standard on archival ink, thinking that it would be a great idea to use such pens for my intricate drawings. Turns out the drawings, some of which I actually
324.
by smokel 1y ago
What will they find out? That we are humans?
325.
by smokel 1y ago
Unfortunately it is not solvable. See this blog post on how to programmatically check for solvability: https://datawookie.dev/blog/2019/04/sliding-puzzle-solvable/
326.
by smokel 1y ago
Perhaps I'm doing something wrong, but the line widths in the characters are quite inconsistent. For example the two l's in "install" have different widths. This makes something pretty amazing look slightly disappointi
327.
by smokel 1y ago
They probably mean CARTO, formerly known as CartoDB. https://carto.com/
328.
by smokel 1y ago
Simply add images and video, and these estimates start to sound like the "640 KB should be enough for everyone". After that, make the robots explore and interact with the world by themselves, to fetch even more data. In all seriou
329.
by smokel 1y ago
The people running AI companies have figured out that there is no such thing as hell. We have to come up with new reasons for people to behave in a friendly way.
330.
by smokel 1y ago
You aren't hiding it, but the announcement and first paragraphs in the README suggest otherwise. It comes off as a bit disingenuous, perhaps you are not aware of this?

More ›