His point is reasonable. And look at programmers (and to a certain extent admins), they do nothing else than automating themselves away. They just know, they are still needed, but at a higher level. You may lose the job as assembler programmer when someone invents a compiler, but you get a job as C++ programmer. You lose the job to implement textbook algorithms when someone licenses a library with the algorithms, but you get the job to build the cool application using the algorithms. Don't fall for the "lump of labor" fallacy.

Who would have thought that anonymous currencies can be used to move money without people knowing where it is coming from and where it is going to?

It doesn't happen YET.

Of course they filter their dataset for pure spam and as best as they can for ads. Training itself also leads away from fitting to certain ads organically (if you want so the model learns what's relevant and whats noise). But wait for spammers starting to optimize to get into training. It's probably not simple, but there is a lot of money in ads and spam, so people will start doing it. You can already pay people to optimize your website for AI bots (with hard to verify results), optimizing spam for AI bots is just the next step.

No

Good syntax + good performance?

I mean look at python used for ML purposes. Filling an array the python way is 100-1000 times slower than using numpy correctly. And guess what? Numpy is faster than writing C. Why? Because the numpy/scipy people use C backends, but not your efficient C code, but heavily optimized code. So while using a simpler syntax you still beat the more efficient language, because the modules of the language you use are implemented in the efficient language by experts who know how to optimize these operations.

Counter argument: If not everyone learns (every subject) the same, not everyone needs to be able to graduate in every subject.
If you have discalculia, don't ask for extra time in your math exams, but choose another subject to study. You won't be happy in a math-heavy job afterward anyway. This also goes for people who only do not like math but may be able to push through ... if you want to challenge yourself do so, but if you struggle don't ask for easy mode but reconsider what's your own strengths.

I don't think as long as Meta Glasses are legal, that you can argue that it is inaccaptable duress to be near someone wearing them. One can criticize Meta, the concept, or the overall idea, but that doesn't justify vigilance.

Then it should be easy to sue her to pay for new Meta Smart glasses. No matter how much she's hailed, the man can get new glasses and laugh about her.

Perplexity is mostly:
- Ask the LLM if it can write an introduction
- Do a bing search
- Ask the LLM what webpages are relevant
- Fetch relevant pages
- Ask the LLM to write an answer
- Ask the LLM to attach sources to its claims
- Ask the LLM to write a conclusion

The answer contains a bit of information from every source but usually not verbatim. I'm also pretty sure that it is instructed not to use longer verbatim quotes, so it does not infringe copyright. On the other hand, it attributes the information to the sources, such that one cannot say it would plagiarize either. Overall the results are pretty solid using the approach and it is particularly good for finding things you vaguely remember like "I think there was a song with a line like ... but I cannot find it. It was rock and people hated it first then everyone loved it" what doesn't work with Google. It also likes to build tables if you say like "I have two options to upgrade my pc ... which has which advantages" and sources both specifications and user opinions from relevant fora, blog articles, etc..

In particular in image processing tasks AI usually beats humans. It's also not a new field, there is nobody saying "AI is hype let's do cancer detection with it" but they are doing cancer detection with it because it is well researched and proven to work.

Machine learning is a subset of AI. AI is the category that even contains the path planning algorithms in your DOS games.

https://ancillary-proxy.atarimworker.io?url=https%3A%2F%2Fpressbooks.ulib.csuohi...

Maybe at least try to read RTFA when people present new developments instead of insisting that the old version did not work so the new cannot work.

I always wonder with these materials, how much heat they accumulate. If it absorbs 99% of light, does it warm up much when hit by average light? I think most material are absorbing like single digit percentages only?

Why do you think they put them out of business? That was never the plan.

Because NYT is not playing fair. They open the paywall to bots to get listed so you have unusable Google results (and a small percentage may sign up instead of clicking the next result) and AI scrapers are run from the same IP ranges as search engine bots. If they wanted a secure paywall every serious web developer would know how to build a secure login system. They actively invest work to build a semi-open paywall page.