Good article overall. Quite often we are caught in our own dilemma and risk being toxic. My only criticism of the article will be not mentioning the risks of being "too empathic" and absorbing other people's problems into yourself. You should not only be able to get into their shoes, but also get out of there too and do it relatively quickly. Otherwise, you cannot survive as a leader of many and will be quickly crushed emotionally.
It depends. If fed with high quality content (e.g. publications) results are decent and mostly on the positive side of time savings/value. Try to ask it a simple question, such as "pros and cons of these two smartphones" and it will suck up marketing bullshit from the internet and give you a convincing, but useless answer. I was really surprised to find this out, because I've been mostly living in the "scientific research" bubble, where it performs good to excellent and really saves time.
Currently the going annual salary for a senior developer in Bulgaria (Eastern Europe) is somewhere between EUR 50k - 75 EUR net, with 25-30 days of paid time off. The gross will be about 25% more (income tax is 10%, social security around 40%, but there is a cap on the salary base, so effectively it is less). For freelance senior developer, the (B2B) daily rate is in the range EUR 350 to EUR 550, depending on the technology and project complexity. Of course, there are rare examples outside of these ranges (e.g. top talent in ML, BigData etc). Leadership roles are usually at 10-20% premium.
The explanation is that they wanted the (perceived) productivity back. Alternatively, they may have wanted to push overall compensation down (e.g. having some employees quit and only luring back the ones they actually like). Of course, it varies a lot from market to market, depending on how hard it is to reduce headcount in the first place (e.g., in Germany, it is much harder than in the US).
but even then it is not so trivial. Yesterday I gave DeepSeek a simple diophantine equation and it got it wrong 3 times, tried to correct itself and didn't end on a correct solution, but rather lied that the final solution is correct.
And what are they going to sell? The weights and the model architecture are already open source. I doubt the datasets of DeepSeek are better than OpenAI's
As others have noted, if one company agrees to the ToS, asks "the right" questions and then publishes the ChatGPT answers, there is not violation of ToS. Then a second company scrapes the published Q&A, along with other information from the internet and again there is no violation (not more than the violations of OpenAI).
In quite many places, hiring consultants has a very high corruption potential (e.g. the hiring manager favoring one of several suppliers). With employees they don't have this leverage.
No need to unionize and strike in this case. Just convince people to stop using Meta's products. I've never had them (excl. a short period of WhatsApp) and never missed them.
Can you be more specific on the use cases that comprise that? E.g. is it driverless company fleets, enabling elderly to drive, making it possible to e.g. sleep while driving, being fancy etc?
It's not the cloud cost that is stopping people from doing this. It is the installation and administration cost. That requires time and expertise that most either don't have or are not willing to spend.
"No Kubernetes because the idle cluster cost with nothing in it is larger than the total cost of my current setup" has been the case for quite a while. I wonder whether it is anywhere on the k8s priority list to fix this. Even K3s consumes a lot more than say docker swarm for more or less the same job.