I read this particular comment[0] and I can't say that I understand the connection you're suggesting. If I ran a company and a country was directly attacking the city were ~1k of my employees live[1], I wouldn't want to continue to do business with them either.
No, I don’t see how one follows from the other. Most of us are not Russian and don’t have websites NC would consider relevant to the war.
On the other hand, most of us probably can imagine not checking our email for three days, getting caught up in some automated purgatory, and experiencing customer neglect. We can imagine this because it is the default customer experience these days.
Well you know I don't see how 'registered in Russia' has anything to do with the war. In 2022 alone about 1.5m people emigrated, and are still 'registered in Russia'.
This was a poorly thought-out virtue-signalling knee-jerk back then and of course everyone in their right mind transferred their domains out because this is not how you do business.
Next thing, you're going to argue that when the US sanctioned the whole population of Crimea back in 2014, that also was "a poorly thought-out virtue-signalling knee-jerk" reaction that worsened, for no good reason, the situation of the helpless Ukrainians who had happened to live there and now also had to live under Russia's brutal rule and face the international isolation. But you'd be wrong.
Are you trying to suggest that the issue in the current post has the same underlying cause (not doing business with countries on the other side of a war)?
I assume that the underlying reason is "badly organized, badly managed, making stupid decisions overall". Certainly sounds like something you should not trust as your business partner.
It has very poor character coverage, especially compared to Iosevka whose author speaks Mandarin as his main language (IIRC), and put a lot of effort into supporting as many languages as possible.
If running systemd, wrap that crap into a cgroup and force kill the whole process tree on exit. I've been doing it for years with badly behaving applications. On phone, from memory, may not be exactly what is needed:
This doesn't really solve the problem, but the good news is jxl has excellent generation loss in lossy mode compared to other formats, if you recompress it with high quality, it keeps original quality pretty well.
Depends on who you listen to, or what you test it on. It goes either way, and avif is typically significantly better at small files sizes where loading speed (or saving data) is the priority.
I am in two opposed minds on this. On the one hand, you're correct. Their job will be made redundant in a couple of years. Some senior people might be retained to orchestrate all the agents, but eventually they will also be replaced. And it's coming for all white collar jobs, not just coders. That's bad.
On the other hand, unlimited intelligence democratises products and services in a revolutionary way. Everyone will be able to do whatever they like in digital space for basically free. No more purchasing software. Make any app/company/website/blog/movie/show/game/forum you can think of. This will quickly bleed into the physical space, with real-world products and services becoming logarithmically cheaper. A true future of abundance.
The big problem is going to be our social and economic transition. Are we going to wait until unemployment is that 10% before we act? 20%? How bad are we going to allow things to become before we implement universal high income? I say universal high income because white collar workers are not going to accept being relegated to an income of $1000/month forever. To achieve universal high income requires eye-watering taxes on the major benefactors of our new economy. They won't accept that without a fight, and it our current fractured political climate, what would it take for everyone to agree on a unified direction?
I am confused about the whole situation let me explain.
It seems like all software will get replaced it by local versions. Suppose I am an individual or a business that was using a SaaS product (such as JIRA).
Let's be generous and imagine that the monthly payout was $20/month for that software (usually for businesses, it is much higher) and one fine morning I login and point a multi modal AI agent to it and clone the whole thing for my local use in let's say 3 months worth of $20/month AI subscription.
Now what? Gradually, that software company would go out of business but that user won't be subscribing forever either to AI inference service and the only constant customer of AI inference would have been the software company which went out of business so... No software company, no AI inference business either?
If I understand your question correctly, it reduces the value of software to almost zero. Meaning all of the SaaS companies are going to have to work very hard to diversify their revenue streams into value added services. That's going to be very difficult or impossible for most to navigate. Hence the term in the industry "SaaSpocalypse."
Some companies will retain a moat. For example, YouTube holds incredible IP, and until AI can replicate that wealth of content at scale for cheap, it will remain profitable. Facebook has the user critical mass. Increasingly, hardware for inference is going to be a moat. So is electricity.
Which loops back to my comment above: this breaks a lot of the market dynamics required for a functioning economy. Accelerationists are convinced that we'll invent new jobs to replace the old jobs, but I don't think that's going to happen. Certainly not at the pace of change we're seeing.
Basically only moat left are either if you have "Data" or "Compute". Software layer is worthless and pretty much reproducible.
Data as in YouTube is all just data. Building a website and app around it isn't the hard problem. Anyone can launch another YouTube in few months but that would be pointless.
Yes, and increasingly, data will no longer be a moat either. These models will be capable of creating all the data/content we could ever want. There will always be a market for "organic" movies with real actors, but we should expect the value of that to decline precipitously.
My investment thesis is currently around these strata:
1. Electricity. All analysis shows massive shortages beginning now with no end in sight. Western governments in particular have massively dropped the ball on this one.
2. Chip design and production. Nvidia and TSMC dominate here for a few reasons. Partners required in the supply chain like ASML benefit, but they're not really the bottleneck right now. We're seeing a lot of investment pouring into this space and we'll see competent design competition (especially from Google's TPUs, and potentially from Apple and AMD). Chip fab is very difficult to get right, and takes a long time to ramp up. See the NAND space.
3. AI labs. These guys are doing to get commoditised. They're trying very hard to prevent distillation but I think it's impossible. The open weight models are not far behind, and the Chinese labs are getting very creative about how they work with limited hardware. If anything, they'll decrease inference cost over time, reducing the value of existing data centers (of which OpenAI and Anthropic have purchased an enormous amount). I am very bearish on the future value of OpenAI and Anthropic.
4. Software. So much software. Everyone gets software. The companies which benefit from this will be able to monetise it. Google, for example, might put ads into VERY capable and free AI. People might accept that. Microsoft has been investing heavily into their data centers, and maybe that's their revenue center in the future. Especially securely hosted services. I expect massive growth in this space in the short term. As long as people still have high paying jobs, they'll pay for well made software and tailored services.
> As long as people still have high paying jobs, they'll pay for well made software and tailored services.
Here's the question. Which high paying jobs will remain? People will still be needed, obviously, but salaries would be in a race to the bottom without some form of govt intervention.
> On the other hand, unlimited intelligence democratises products and services in a revolutionary way. Everyone will be able to do whatever they like in digital space for basically free. No more purchasing software. Make any app/company/website/blog/movie/show/game/forum you can think of. This will quickly bleed into the physical space, with real-world products and services becoming logarithmically cheaper. A true future of abundance.
This isn't happening yet, SaaS is only growing more in the era of AI. Anecdotally, my own small SaaS has only grown instead of shrunk as well.
The name "universal high income" should be tipping you off to the issue with it. If money has less importance then what tools will individuals have to change their station? How will anyone other than the permanently entrenched rulers affect change? We don't have a good system now, but it seems like we're hell bent on making it even worse.
Edit: UBI being pushed by asset owners is a step towards Blade Runner rather than Star Trek.
I'd like to see AI "democratize" the world, but I have a much more pessimistic outlook. One thing we see with technological revolutions is that (once the dust settles) they provide general improvements for all strata of societies. But as some boundaries are erased, others are magnified a hundredfold wherever robber barons can stake an exclusive claim.
If genuine AGI is developed, its everyday conveniences may continue to be shared with the common masses, but the powers that own the technology and the data centers will have no incentive to give everyone access equal to superintelligence, and that includes giving governments the same access.
Every stage of the industrial age has seen powerful, wealthy people/groups (whether you call them oligarchs or industrialists or whatever) become able to exert disproportionate power over the state, and we're already in a place where this is happening. Humans have shown time and again that once you have some resource (whether it be oil, gold, or data centers), someone will want to own all of it, and that they're always willing fight for that ownership, and this fight rarely benefits ordinary humans like us.
Developed countries have been enjoying an unusually long peace time thanks to the wonders of capitalism and the emergence of stable, progressive capitalist-friendly democracies. The road to utopian basic income and limitless leisure is going to be fraught with riots and violence, and make Marx suddenly utterly relevant once again, because there's too much at stake, and too stark a division between the haves and the have-nots.
The alternative timeline is one where governments get together to regulate AI into some kind of international agreement under the UN. That requires taking power away from the private corporations; historically, that's been a tough sell.
I believe that the middle ground for this is to keep all these jobs orchestrating agents or whatever the AI future invents for us, but with lower wages. The golden era of dev wages will be gone. There is no incentive for companies paying people 100-500k-1m to write prompts and debug agents, independent of their level.
This might be how it shakes out but I don't think it's acceptable, and I don't think white collar workers will think it's acceptable. Taking high income people and making them low income people (usually for life) has a tendency to politically radicalise people. Especially in light of the unprecedented productivity and profits which businesses will experience. People will demand a slice of the pie.
One kind of intervention that could work would be licensing, like with some engineering professions. Make it illegal to work on software without a professional license; same as with doctors. Engineers would be free to work with AI, but the final responsibility would fall on them.
This artificial control works to prevent oversupply and protect salaries. Unfortunately, the businesses calling the shots won't ever allow this to happen.
So? They'll be declared domestic terrorists and that'll be that. I mean, you cannot demand something from an entity that doesn't care about you, doesn't need anything from you, and over which you have no leverage.
I would say, be afraid and use the fear (not to ruminate, but to organise and do things), but also be optimistic, and use that, too (not to imagine what people will do then when they're not doing it now, but to believe it's possible helping making it reality, now).
There is no credible path towards stopping progress on AI. Nobody knows exactly how this will play out. But whichever way it turns out, being an skilled AI practitioner will give you strictly more options than not being one.
Because (maybe particularly) at 1e100 scale, having a super intelligent butler that does whatever everyone asks of it can easily get you to hundreds of thousands of daily PRs merged to the monorepo, but won't necessarily help coordinate efforts, and may actually cause certain projects to collapse under their own weight.
My believe is that current LLMs performs very well in the constraint environment with lots of data available, e.g. we have lots of code on github, public discussion with feedbacks, compiler and tests failures as feedback, so LLMs can train on trillions data records.
That's not true for many unconstrained real world scenarios, and LLMs currently perform poorly there. Another breakthrough is needed to overcome this.
Finally circumstances will force me to become a goat farmer, a desire I've always secretly aspired to but never felt comfortable expressing, let alone acting upon
Now we will finally be free, without having to make hard decisions. It will be decided for us.
So basically people who didn't, or couldn't prep for this will be 'effed. Not accusing you or something. Just thinking out aloud. I am/will be one of those people.
I actually have complex feelings about it as well.
The way I see it is that history is going to advance whether you like it or not and you might as well ride the wave.
I understand I'm lucky to be insulated from the crisis a bit, though. I do think once SWEs lose their jobs every other office worker is going to follow, so we're in the same boat.
I see. Thorughout the history there has been no such looming danger of mass unemployment looming and what comes with that. But I do also feel the same - eventually we will be in it together. And if that mass scale unemployment comes to pass, then it may lead to a bit of interesting times so to speak. But yeah, that insulation does make watching all this unfold in real time a bit less uncomfortable ;-)
Sometimes I wonder AI won't have to hack military drones to end humans en masse. Making relevance, employment, and other life conditions for most so pathetic that it will start making it happen on its own.
you do need a precise prompt to do what it's asked and there are 50-100 things that needs to work to meet the criteria. Little things need to work, and work well
As someone who grew up poor, and then used to be on food stamps and other state aid as a young adult, that's a particularly pesky problem. Dread is a more apt feeling than joy or elation in that situation.
Out of sight, out of mind. It is unexciting because in that future, there is nothing to think about related to this topic. What is exciting about something that is irrelevant for humans since nobody looks at it anymore?
https://news.ycombinator.com/item?id=30504812
(Before reacting to the headline, please read at least some of the discussion.)