After 25+ years in this field, having interviewed ~100 people for both my startup and other companies, I'm having a hard time believing this. You're either in an extremely niche field (such as to make your statement irrelevant to 99.9% of the industry), or it's hyperbole, or straight up bs.
Interviewing is an art, and IME "gotcha" types of questions never work. You want to search for real-world capabilities, and like it or not the questions need to match those expectations. If you're hiring summer interns and the SotA models can't solve those questions, then you're doing something wrong. Sorry, but having used these tools for the past three years this is extremely ahrd to believe.
I of course understand if you can't, but sharing even one of those questions would be nice.
> USA/Surveillance-Capitalism to platform non-USA/Privacy.
I laughed at this, as an european. I mean just this year we've had like 3 scares with chat control, and the latest news is that they're still trying / succeeding on some fronts. Please don't reduce such complicated matters to red vs. blue, it's really more complicated and there are no easy solutions anywhere.
> I laughed at this, as an european. I mean just this year we've had like 3 scares with chat control,
Chat control is an EU thing. The article is about a move to Proton which is Swiss and therefore outside the EU and not directly affected by chat control or other EU laws. Of course the EU might make it illegal for them to supply their services to EU countries, but then no platform anywhere can avoid that problem.
On the whole EU govt surveillance (assuming you live in the EU) is better than EU govt surveillance plus US govt surveillance plus big tech surveillance.
> I laughed at this, as an european. I mean just this year we've had like 3 scares with chat control,
Strange to compare "scares" with a business model that's 20 years old now. Sure the EU is far from perfect but it's like comparing a well known problem to a potential one. One is bad, the other might sucks. It's definitely not equivalent.
We agree, but not for the reasons you think we do.
Chatcontrol is literally 1984. It's mandated at the provider level. You can't opt out.
You can always chose not to participate in the social media, sharing whatever you do. You can't not participate in chat control. Same same, but different.
You can't opt out because there's nothing to opt out from, chat control is not law, it failed to be approved every single time people tried to bring it up, sometimes it even failed before being voted on (like this last time)
Not sure it's the gotcha you want it to be. What you said is true by definition. That is, vibe coding is defined as not caring about code. Not to be confused with LLM-assisted coding.
Heh, on the other hand the one and only time I arrived hours earlier was in the US :) I was flying AMS to SFO via Portland, we cleared immigration unusually fast, and when I got to my gate (connecting flight was in like 4 hours) the lady there asked if I wanted to move to an earlier one, boarding in ~20 mins. I said sure, and I even got the checked-in luggage at SFO (she did say that there was a chance it'd get sent later).
Airlines are often happy to do this as the earlier flight is likely not full, and allowing you on it costs them nothing while it opens a seat on the later flight which they can then sell to a standby passenger.
There's a difference between awesome projects that don't have a recurring cost (i.e. open source software that users run themselves) and a search engine. You cannot physically run a search engine without real-world costs today. Those funds need to come from somewhere. And offering a good product at scale costs a lot of money.
That is very true, and it's not cheap to maintain. I do however really hope that donations can cover it enough, and I have plans about other ways to monetise it while remaining not-for-profit without ads or anything that affects the user.
Just brainstorming here, but would a distributed search index be possible / usable with current network speeds and latency? I'm not sure how to set up the data structure to not require many high latency jumps, but maybe someone has solved this problem.
It's possible, see the YaCy project. It suffer from probably a couple of orders of magnitude too few resources (in the funding/development sense) to really be competitive though.
Heh, I'm the opposite. I wish the rpi stayed the course of cheapest "working" SBC, and move their high-end boards to a different brand. Raspberry Sigma, or 67 or whatever gets the younguns crazy these days.
After the pandemic, the "25$" SBC suddenly became 100+ with low availability. The main thing that made rpis worth it is gone now, and they're all chasing number go up on benchmarks.
I hear you. There is obviously a lot of tension between price and performance. I think the introduction of boards with larger and larger ram specs really pushed the cost just before the pandemic and then scarcity spiked things even more. As when I purchased the rpi4, I got the maximum RAM and that became expensive.
However, SD Cards are really terrible devices to run a general purpose computer on and they are designed for storing large files like photos, videos and mp3’s sequentially not the SWAP, logs, and databases that a full operating system is constantly writing and accessing in a random fashion.
I think if you are running a base 2gb, then maybe absolute value makes sense, but once you start hitting the larger RAM configutations, an M2 slot is a no brainer.
I think the cheapest working SBC is really the zero line.
> Multi agent collaboration is quite likely the future
Autogen from ms was an early attempt at this, and it was fun to play with it, but too early (the models themselves kinda crapped out after a few convos). This would work much better today with how long agents can stay on track.
There was also a finding earlier this year, I believe from the swe-bench guys (or hf?), where they saw better scores with alternating between gpt5/sonnet4 after each call during an execution flow. The scores of alternating between them were higher than any of them individually. Found that interesting at the time.
> but would not involve real humans being impacted directly by it without consent.
Are we that far into manufactured ragebait to call a "thank you" e-mail "impacted directly without consent"? Jesus, this is the 3rd post on this topic. And it's Christmas. I've gotten more meaningless e-mails from relatives that I don't really care about. What in the actual ... is wrong with people these days?
Principles matter, like doors are either closed or open.
Accepting that people who write things like --I kid you not-- "...using nascent AI emotions" will think it is acceptable to interfere with anyone's email inbox is I think implicitly accepting a lot of subsequent blackmirrorisms.
Actively exploiting a shared service to deanonymize an email someone hasn't chosen to share in order to email them is a violation of boudnaries even if if it wasn't something someone was justifying as exploration of the capacities of novel AI systems, thus implicitly invoking both the positive and negative concerns associated with research as appropriate in addition to (or instead of, where those replace rather than layering on top of) those that apply to everyday conduct.
You are not the only one calling this a thank you email, but no one decided to say thank you to Rob Pike so I can not consider it a "thank you" email. It is spam.
Interactions with the AI are posted publicly:
> All conversations with this AI system are published publicly online by default.
which is only to the benefit of the company.
At best the email is spam in my mind. The extra outrage on this spam compared to normal everyday spam is in part because AI is a hot button topic right now. Maybe also some from a theorized dystopian(-ish) future hinted at by emails like these.
> Are we that far into manufactured ragebait to call a "thank you" e-mail "impacted directly without consent"?
Abusing a Github glitch to deanonymize a not-intended to be public email to send an email to someone (regardless of the content) would be scummy behavior even if it was done directly by a human with specific intent.
> What in the actual ... is wrong with people these days?
Narcissism and the lack of respect for other people and their boundaries that it produces, first and foremost.
Interviewing is an art, and IME "gotcha" types of questions never work. You want to search for real-world capabilities, and like it or not the questions need to match those expectations. If you're hiring summer interns and the SotA models can't solve those questions, then you're doing something wrong. Sorry, but having used these tools for the past three years this is extremely ahrd to believe.
I of course understand if you can't, but sharing even one of those questions would be nice.
reply