@Da7_Tech They are extremely smart + talented researchers, but they’re massively handicapping Chinese AI 10y out by taking the easy way out. But also, compute scarcity. Even the most recent Chinese AI raise was measly compared to some unicorn raises.
@forefy i was doing this exact task last night during an engagement and honestly for real work id rather use daybreak red or a open model on ec2 instead of blindsiding a frontier model. if you have to exfil anything (esp sensitive), forget it altogether
@Akintola_steve AWS, GCP, Azure are all hyperscalers in the way they are massively scaled and financed + paid off compute cheaper than everyone else. Im no AWS shill but i dislike seeing people think they can charge for a managed-managed EC2 w/ markup
@manonsbat@s0menumber5@sadprintboi yes we would. do you think mRNA research is in any way attributable to them? scientists developing a highly processed chemical with a plant being one of the many inputs does not mean modern medicine is downstream from what is essentially a placebo application
@Questionvibing Honestly i don’t hate this guy anymore at all. He been dealt such a horrible hand but made the most of it and he deserves to live in peace now unbothered. I don’t place anything done under his admin on him, he was not mentally fit. In retrospect, pretty unmemorable term
@jmrphy Yes but some good advice is to never get yourslef into a situation where you desperately need AWS support. They literally just will not respond, maybe after 2-3 weeks, but by then its over.
@mschoening@Mappletons@Tailscale Im not even joking I had a similar thought that eventually everyone is going to have to go back to wired intranets. Tailscale is not low enough though. You can throw compute at anything now and brute force it given enough time, so I only see a hard physical separation working.
@wolfofbaystreet NGL throwing rocks is bad overall. When you outgrow it you will absolutely be called out for it there is a long list of rock throwers who failed because of this