> would we trust a Software engineer, which in general don’t always obtain the mathematical foundation to understand deep learning in the first place, on the trajectory of AI?
Valid point, but it suggests a mathematician who understands the math behind AI is more capable of grasping its trajectory, which is probably not the case.
People who are deep in the inner workings of this stuff day in and day out are the only ones who have a chance at having any real insight.
I think more broadly that grabbing attention with predictions and hot takes has become lucrative, and we definitely don’t celebrate prediction accuracy.
> Websites are still inferior to Flash of the early 2000s. It's taken decades and they can only mimic a fraction of its power.
Is this a troll? What could an application do with Flash in 2005 that we can't do with a modern web application today (excluding the obvious answer of runtime vulnerabilities that allowed apps to escape the sandbox)?
> What could an application do with Flash in 2005 that we can't do with a modern web application today
Show me the JavaScript framework (or tool that exports JS) that you can give it to a middle schooler and have them make a cartoon with audio and moving images that they can draw themselves, while responding to user input. Have the exported artifact be consistent across all major operating systems and browsers.
Flash had its problems but as a user, it looked sharper and smoother than even current websites. And its editor gave non-tech users ability to create amazing animations, interfaces, and even games.
wasn't some of that smoothness because it ran at a 100hz tick without any way of adapting it (and still running existing code)? That was the complaint I kept hearing from people attempting to make flash on phones viable (this led to ludicrous battery consumption)
That’s not at all how I remember it (having used it in the Macromedia Flash 8 era, around 2006 or so). You would set your animation’s framerate and that would determine how often your `onEnterFrame` would run.
This was intended for a slow 2004-era computer with a 4x3 (probably 1024x768) display, where it worked very well.
But it's not 2004 anymore; things are much faster and screens have gotten a lot bigger. Here in 2026, Salad Fingers renders out fine at higher resolutions, and at different aspect ratios. It works great on my desktop at 1080p, without stretching [and with some probably-unintentional extra content on the sides]. It even works on my pocket supercomputer's 3200x1440 20:9 display.
Vectors are fun, and they scale as technology improves. The lines remain smooth and defined. And with Flash, that's a built-in: An unaltered 22-year-old digital animation still looks crisp.
For contrast, if Salad Fingers had been published on YouTube way back around that time, it would have been in chonky fixed-pixel 320x240. Maybe that would be as good as it would ever get unless it were rendered and uploaded at higher resolutions later.
Conceptually maybe you can compile flash to SVG+js but this has nothing to do with the point. Many insist (I have no direct experience) that the flash ecosystem (especially the editor) was and is unsurpassed as a publishing platform for interactive experiences.
Today with the current focus on mobile+low latency+e-commerce optimizations flash would probably have shown a lot of limitations, yet JavaScript, SVG, canvas, http webgl etc still fail to provide a "competitor" to what flash used to be.
The web simply went in a different direction, one that left many unsatisfied
The narrow point I was addressing was the claim "it looked sharper and smoother than even current websites".
SVGA graphics, being vector-based (as the name suggests), are indeed sharp, most notably when scaling up or down, and I've encountered SVGA-based interactive graphics which are reminiscent of Flash-based animations in that specific regard.
Again, I'm not addressing other aspects of these options, and I do very little direct development of this type.
I'm quite familiar with the claims that Flash was attractive to publishers and creators. On the receiving end, I was less impressed. Odd Todd excepted. The proprietary nature (I generally run Linux) and constant security concerns, as well as hiding web content within an inaccessible format (e.g., text couldn't be readily extracted) were all frustrations. I'm also generally not a fan of any animation.
what’s he is referring is the editor and the easy way of drawing things, still agree we can do things today but a easy to draw editor like that is missing. I was a fan of flash and fireworks.
The editor was a scripted timeline, similar to a video or animation timeline. It was fantastic for creatives, but counterintuitive for programmers, so most devs hated it.
This went away not only because flash died, but also as the internet commercialized.
I mean, consider this: McDonald’s used to be fun and colorful. Now every McDonald’s is boring and gray. And, wait, every store is boring and gray! And flash had nothing to do with that.
But McDonald's has been commercialized since forever, as have those other stores. Something seems to have changed in the culture where businesses only want to offer bland, boring experiences any more. It can't be just the profit motive, because that always existed.
What was possible in Flash in 2005 in IE6 on a Duron processor with 32MB RAM is possible today in a gargantuan bloatware browser that eats all your hardware.
Probably because AI coding has only worked at all for a couple years and has only gotten good in like the last year?
The rate of improvement has been fast. Maybe it’ll plateau soon, or maybe we’ll have LLMs improving themselves rapidly. At this point it’s too early to say.
I don’t remember where I heard it, but there’s a saying that people overestimate how much can be accomplished in a year and underestimate how much can be accomplished in 10 years.
If we get to 2030 and still people are wondering where the breakthrough is, then I think I’d be agreeing with your skepticism. But I just think it’s too early to judge that yet.
on what? Who the fuck would go full transparency of what's in their black box in this hostile culture of AI hatred? None of us can put a number on what code we've used in our services that was written by humans and long may it last.
They literally can’t go full-transparency. I know a high-level insider, and the fact is that even the folks implementing things don’t actually know how it works, only that it does, and how to get it to generally behave.
N=1, but Claude etc. have made a huge difference to my life personally.
Built a bunch of software tools to streamline my small ecommerce business - while also running it - and things have turned around from "losing money and ready to pull the plug" to "looking at our best financial year on record" in the span of about 8 months.
I could imagine it wouldn't make a huge difference to the life of someone deeply entrenched in a traditional tech role, trying to get an extra 9 of reliability in a service or roll out a new carefully planned and QA'd feature.
But for tech-adjacent people, it gives us something "good enough", instantly, and basically for free.
That doesn't include the other things I've got it to do (gave Claude SSH access and got it to successfully debug a hang on my Ubuntu server, chucked Codex in a folder full of financial data and got it to find every piece of misclassified payroll transaction data)
Genuinely the biggest breakthrough for "casual" tech users since Excel.
The joke used to be “be nice or I’ll replace you with a small shell script” - Claude lets you actually get those scripts written which often aren’t replacing anyone but are automating away part of the daily hassles.
why do you think so? they provide some evidence of this in the article, but there have been several improvements in e.g. nanogpt-speedrun or openai parameter golf made by AIs
It's pretty crazy that a company like Anthropic no longer needs to hire Software Engineers, because their software engineers itself. If that's not a break through I don't know what is!
edit: it looks like I was wrong and they're still hiring many software engineers. Not completely sure why that is just yet.
> Having an account on a company's platform is a privilege, not a right.
Businesses can lose a lot traffic by not being present on Facebook and Instagram, so being unjustifiably banned is doing measurable financial harm in many cases.
Even as an individual it can be a huge pain to not have Facebook. The local individual sales market (e.g. classified ads) is dominated by Facebook Marketplace now, for example, and not having access to that market makes it difficult to sell things.
Meta has a responsibility to the community because of their position as the de facto platform for many activities. They've even intentionally positioned themselves to dominate. Having laws requiring them to act responsibly is totally justifiable.
It's not Meta's - or any other company's - responsibility to care about the impact on users having or not having access to their platform, beyond how it affects them financially. What you're saying is that, because they're a huge company, they MUST - be forced to - allow access to anyone who might be somehow "disadvantaged" by not having access to their platform. I'd say that's ludicrous. The only responsibility Meta has is to the interests of its shareholders, which is primarily to grow their investment.
Regulatory laws can always be made of course, but it's unrealistic to expect that Meta, or any other company, will do any more than is required to ensure they're turning a profit.
Yeah it's bad, but AI isn't required for this type of thing to work.
My anecdotal experience is my Facebook account was compromised several years ago after TOTP 2FA was disabled. Didn't exactly give me a warm fuzzy about Facebook security policies at the time, and this new attack just reaffirms that.
I'm not a handyman, but I am a man who happens to be handy.
I have done quite a bit of painting and caulking for a guy who's not in the profession. I despise both with a passion, though, especially caulking, and I have never once been satisfied with a single paint or caulk job I've done. I feel like I'm the embodiment of "be bad at this for a long time," although I'm objectively probably halfway decent at it.
That is to say I think Ira Glass' quote of "You've just gotta fight your way through" to get where you want to be seems especially meaningful in the context of something like painting, where most everyone _can_ do it (or writing / storytelling in Ira's case), but very few are actually good at it.
You need a silicone caulking tool, and a video. I have spent many years caulking like a fool listening to other fools who spray water and use their thumbs. Don't. Use the tool. Use the kind with a little oval tip usually (I mean, there are exceptions with harder caulks but for softer e.g bathroom caulks this is more superior.)
There's one UK guy on YouTube that convinced me of the evils of water/iso sprays and the beauty of the proper silicone caulking tool.
The little wedge shaped caulking tools btw are not enough, as you need some stick to it so you can get around certain angles/items.
I don't know if it's specifically UK terminology, but here we don't call bathroom silicone "caulk" - we use that term for decorators caulk, which is much more sticky; so needs to be treated differently to silicone. Definitely true that the little tools are amazing for bathroom silicone, not sure what you do with the other stuff
True or mostly true could easily be argued from a statistical likelihood perspective: life exists on Earth and, based on what we know, Earth doesn't appear to be all that special in a very large universe.
I think you could come up with a reasonable argument for any of the responses, hence the problem with the methodology.
> It's a weird fact claim, because the ground truth is "nobody knows for sure" and that's not one of the available options.
It's even weirder to suggest that the disagreement is indicative of a problem. If you asked five very knowledgeable humans on this subject to select the correct answer on a multiple-choice questionnaire, they would almost certainly vary significantly more than these 5 LLMs.
Not to say that hallucination isn't a problem, but this is a lousy way to test it.
What are you talking about, it had the option for nuanced responses, but it chose the more binary responses. It could have chosen no explanations, no qualifiers but instead it showed off LLMs incapability for nuance.
These types of experiments prove to me that there is no real "reasoning" happening and "reasoning/thinking" tokens as a concept are mostly there to convince people to use models that consume more tokens and produce more revenue. The output from reasoning models might be more accurate, but its just a consequence of a longer inference runtime, there is no "reasoning" happening, reasoning is just sales/UX bullsh*t.
Valid point, but it suggests a mathematician who understands the math behind AI is more capable of grasping its trajectory, which is probably not the case.
People who are deep in the inner workings of this stuff day in and day out are the only ones who have a chance at having any real insight.
reply