People get really confused when I skim through a massive amount of AI slop engineering garbage text, and reply with 6 one-line questions.
The machines are getting better quickly, but it's still a challenge to prevent them from overcomplicating projects.
While it will likely not have a systemic effect on bitcoin, from a philosophical and culture level, this is the worst thing to happen to bitcoin in a very long time. Coldcard users were in theory doing everything right. I’m very sad for all affected #coldcard users.
1/ Earlier today, our Bitcoin engineering and security teams at Block began investigating reports of non-Bitkey wallets being drained. To proactively protect our customers, we began investigating immediately. Here’s what we found 🧵
For remote teams, the higher up you are on the org chart, the worse your wifi will be during calls. The corollary is: the higher you are on the org chart, it is less likely that the team will tell you your wifi is horrible and they can't understand you.
On a more serious note, whether or not the AWS billing glitch was AI-related, it's interesting to think about the tradeoff we're now generally accepting towards a society: we're ok with AI making some mistakes (even fatal ones, at the extreme) if the present or future good to society outweighs the harm to an individual. Tech has long accepted move fast and break things as a mantra. The more extreme example is that @Tesla FSD has arguably resulted in some deaths from people who over-relied on it. However, we know that eventually self-driving level 5 will be here, and it will save a million lives a year annually by some estimates.
In the coming years, we'll see the same calculus applied to fields such as medicine and law. It will be rough to be one of the ones where a human would have caught something that AI missed, but the millions (billions?) of people who will have affordable access to advanced medical specialist agents will not want it any other way. We're going to watch the greater-good theory speedrun many areas of life. It's going to be a huge bummer though if the machines start prioritizing their own greater good over humanity's good.
I really enjoyed getting a message from a customer this morning while I was eating my breakfast, asking why they had a $35 BILLION dollar AWS bill. Now taking bets on whether AI was involved in pushing this buggy code to prod. AI is taking move fast and break things to a new level!
📣 Service Update: AWS Billing Console
We are investigating issues with Cost Explorer reflecting inaccurate estimated billing data.
📊 Track status: https://t.co/6qSnkuEzc4
#AWS#AWSSupport
@skylar_b_payne This has not been my experience. Been running 5.5xhigh, and the compaction has behaved so well that I've become very lazy about clearing context, because the auto-management works so well.
Early in my academic career, the teachers would assign a paper with a min page #. In legal writing, many courts now have a max. I think we'll get there with AI in business correspondence. The AIs generate huge walls of text, but aren't always able to discern what's truly important.
It is wild how tech discussions on slack have become engineers pasting huge chunks of AI text back and forth. I hope we find a way to retain critical thinking skills, as a society, instead of outsourcing it all to AI. Steering is not being talked about enough.
@CarrascosaCris_ Yes, regulators really enjoy being told how to do their job in a condescending way, lol. Binance is lucky that they didn’t have a mica license on 10/10.
“Digital credit” is an attempt to convert capital appreciation of bitcoin into dividends. The problem is they didn’t model the volatility of bitcoin correctly, and now the volatility is reflected in $STRC ‘s departure from par.
@__tinygrad__ Love to see it. It’s so easy to take software for granted when the assumption is that you can vibe code up something that “works”. None of it would even be possible without the solid foundations of decades of labor on the building blocks like Linux, python, unix cli tools, etc.