Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
davedx
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
31 ms
·
691.
▲
by
davedx
2y ago
What, no it doesn’t, that’s “AGI” - it has a G in it. This is ML/AI
692.
▲
by
davedx
2y ago
Using OAuth 2.0 [as documented, when the implementation matches the docs] isn’t too bad, but many many companies diverge in weird ways IME.
693.
▲
by
davedx
2y ago
What’s wild about it? This case is about its competitive practices in the marketplace, not privacy
694.
▲
by
davedx
2y ago
“AI has peaked” “AI is a bubble” We’re still scratching the surface of what’s possible. I’m hugely optimistic about the future, in a way I never was in other hype/tech cycles.
695.
▲
by
davedx
2y ago
Where are you getting that price from? The numbers I can find are roughly half that: https://www.uswitch.com/gas-electricity/guides/regional-ener... Please post sources.
696.
▲
by
davedx
2y ago
> there are a lot of perks that come with a full-fledged control plane that would almost certainly be worth the added cost and complexity. Such as? Logging is more complicated with multi container microservice deployments. Deploying is m
697.
▲
by
davedx
2y ago
> Salesforce's pricing feels wrong out of principle What do you mean exactly? If it takes multiple engineers multiple months to build an alternative on kubernetes, then it sounds like Heroku is worth it to a lot of companies. These
698.
▲
by
davedx
2y ago
I love Consider Phlebas, it’s a right old romp, the pace always pulls me right in.
699.
▲
by
davedx
2y ago
The humans are still there, just left to do their thing pottering around on Earth doing the odd genocide. (State of the Art)
700.
▲
by
davedx
2y ago
Deportation operations and camps in the UK: https://www.rescue.org/uk/article/rwanda-plan-explained-why-... https://www.gov.uk/immigration-removal-centre
701.
▲
by
davedx
2y ago
Yeah, isn’t that how the (most recent) opioid epidemic got started - a pharma company telling everyone their new opioid was way less addictive - almost a sort of miracle cure?
702.
▲
by
davedx
2y ago
Sell AMZN and buy LLY
703.
▲
by
davedx
2y ago
Thanks for the considered answer. What is the cortical algorithm? (Yeah, it's been quite a few years since I did any bio psych...)
704.
▲
by
davedx
2y ago
Physically, sure. But 1) feedback (more synapses/backprop) and 2) connectedness (huge complex graphs) of both produce very similar intelligent (or "pseudo-intelligent" if you like) emergent properties. I'm pretty sure 5
705.
▲
by
davedx
2y ago
Sure, it's a model. But don't we think neural networks and human brains are primarily about their connectedness and feedback mechanisms though? (I did AI and Psychology at degree level, I understand there are definitely also big d
706.
▲
by
davedx
2y ago
> Modern ANN architectures are not actually capable of long-term learning What do you think training (and fine-tuning) does?
707.
▲
by
davedx
2y ago
Imagine empowering accountants and all other knowledge workers, on steroids, drastically simplifying all their day to day tasks and reducing them to purely executive functions. Imagine organizing the world's data and knowledge, and int
708.
▲
by
davedx
2y ago
> There is significant possibility that true AI (what Ilia calls superintelligence) is impossible to build using neural networks What evidence can you provide to back up the statement of this "significant possibility"? Human br
709.
▲
by
davedx
2y ago
I worked with Scala professionally once and agree it’s very similar. Honestly what put me off Scala the most were the atrocious compile times, it was like working with big complex C++ projects in that respect. Sorry but my time is more valu
710.
▲
by
davedx
2y ago
Great programming lang article, starts with premises then shows examples, iterations, and final words. Gleam looks really interesting, though personally I’ll stick with TypeScript because it’s “good enough” and I understand where mutability
711.
▲
by
davedx
2y ago
No, I always indicate. I am saying Tesla’s various safety systems frequently have false positives. Ask people who tried to get into the FSD beta program how reliable these systems are.
712.
▲
by
davedx
2y ago
Yeah this is most modern games - bewildering options. It’s the reason I stopped playing the Witcher 3 and various other games, and why Stray was amazing and so much fun to play
713.
▲
by
davedx
2y ago
Distractions are often the safety systems themselves! Over the last 4 years, my Model 3 has progressively started beeping and chiming more and more at me, taking over my steering because it thinks I'm drifting out of my lane when I
714.
▲
by
davedx
2y ago
Wait what. I don’t know if my inverter does what they say. For one thing the vendor went bankrupt so there is no cloud dashboard anymore. For another there are hundreds of inverter vendors not one single one. And I am highly sceptical the b
715.
▲
by
davedx
2y ago
I don’t think so, each battery would be discharging at its rated current? The problem would be what happens “upstream” in the grid
716.
▲
by
davedx
2y ago
Not much about the potential severity of hacking something like this: You could probably blow circuit breakers in MV substations with a vulnerability like this if there are enough batteries under control in the same area and you get them to
717.
▲
by
davedx
2y ago
He has his own brain, some advice based on actual experience is not going to ruin him. Please don’t be so dramatic
718.
▲
by
davedx
2y ago
My freelancer estimate is based on 100/hour, that’s actually quite high for Europe
719.
▲
by
davedx
2y ago
Do you know what a freelancer is? You don’t make tax contributions for them. They bring their own hardware and usually software unless otherwise agreed. The rest of the stuff is just a laundry list you made up to try and blow costs way past
720.
▲
by
davedx
2y ago
IME GPT-4V is a lot better than Tesseract, including on scanned document PDFs. The thing about frontier models is they aren’t free but they keep getting better too. I’m not using tesseract for anything anymore, for my tasks it’s obsolete.
More ›