This is the start of the end of generative AI music.
Without the entire music catalog, they are just not going to be able get the diversity and quality of outputs they once had, and the labels are always going to have their hands out wanting more.
Udio and Suno have demonstrated that okay-quality new music can be created by training on vast collections of pirated music, and, as expected, the music labels shut them down.
I find it useful. A nice little tool in the toolkit: saves a bunch of typing, helps to over come inertia, helps me find things in unfamiliar parts of the codebase, amongst other things.
But for it to be useful, you have to already know what you're doing. You need to tell it where to look. Review what it does carefully. Also, sometimes I find particular hairy bits of code need to be written completely by hand, so I can fully internalise the problem. Only once I've internalised hard parts of codebase can I effectively guide CC. Plus there's so many other things in my day-to-day where next token predictors are just not useful.
In short, its useful but no one's losing a job because it exists. Also, the idea of having non-experts manage software systems at any moderate and above level of complexity is still laughable.
I don't think the concern is that non-experts would manage large software systems, but that experts would use it to manage larger software systems on their own before needing to hire additional devs, and in that way reduce the number of available roles. I.e. it increases the "pain threshold" before I would say to myself "it's worth the hassle to hire and onboard another dev to help with this".
> It's still mind boggling to me that governments didn't say "Don't wear a mask for yourself -- wear one to save your neighbor."
I mean... they did?
Like in the UK's guidance literally the second sentence is "Face coverings are primarily worn to protect others because they cover the nose and mouth, which are the main sources of emission of the virus that causes coronavirus infection (COVID-19)."
The point of masks, originally, was to catch saliva drops from surgeons as they worked over an open body, not to stop viruses.
For COVID its use was novel. But having an intention isn't enough. It must actually work. Otherwise, you are just engaging in witchcraft and tomfoolery.
The respiratory droplet model of how COVID spread was wrong, which was proven by lots of real world evidence. Look at how the Diamond Princess worked out and please explain how that was compatible with either masks or lockdowns working? SARS-CoV-2 spreads like every other respiratory virus, as a gaseous aerosol that doesn't care about masks in the slightest.
I'm not sure where you're getting this from. Repeated studies continue to affirm that COVID is spread by respiratory droplets and that masks are effective in reducing transmission.
Your first link is agreeing with me. "Wang et al. review recent advances ... The authors suggest that airborne transmission may be the dominant form of transmission for several respiratory pathogens, including SARS-CoV-2" (airborne transmission here means not via droplets but gaseous clouds of the type that just go around or through masks). Anyway.
Case study: Diamond Princess was a cruise ship at sea (i.e. sealed environment) in which the SARS-CoV-2 index case passenger was found very early. They locked everyone in their cabins and food was left outside the doors. This was an effective way to eliminate contact.
Implication: If SARS-CoV-2 spread only via respiratory droplets, that strategy would have killed the epidemic in its tracks.
Observed behavior: COVID cases appeared randomly throughout the entire ship without rhyme or reason, until eventually it petered out having sickened many but not all of the people on board.
Conclusion: SARS-CoV-2 is just like SARS-1, infected patients create gaseous aerosols which can spread long distances through air ducts. Exposure on board was 100%, but some people were naturally immune due to having been exposed to similar viral proteins in the past.
External validity: mask wearing and lockdowns routinely went from 0% to 100% to 0% again overnight, and yet no impacts can be seen in derivatives of the case rates. As creating such inflections in the derivative was the sole justification for the policies, these policies failed. Contact tracing studies estimated a maybe 80% pre-existing immunity rate to the initial variants based on lack of household transmission.
Why does the real data conflict with so many post-2020 studies? Check the details and you'll find the ones claiming masks work are all pseudo-scientific. Many of them don't even use real world evidence at all, they're purely modeling exercises in which academics start by assuming masks work and then show a reduction in simulated infection rate! Such papers have no scientific validity, it's merely writing a video game that reflects your own beliefs back at you.
If you narrow it down to just studies that use the scientific method correctly and which are built on strong evidence, no convincing findings remain. That was the conclusion of the Cochrane Review, and it was the scientific consensus before COVID too. You know, when the health officials started out by saying (everywhere, globally) that there was no evidence masks worked, it wasn't really some enormously cunning global conspiracy to limit demand. Use Google Scholar and see for yourself.
The switch to claiming masks worked came about because epidemiologists built models that shocked the world with their predictions of mass death and crisis. Lots of people became so terrified they'd barely leave their homes. Politicians and Public Health needed a way to talk people down from the ledge so this idea of mask mandates came, but it came from the politicians. Politicians: "Do masks work?", PH: "We have no evidence they would", "But would masks hurt?", "Well.... no ... guess they can't really hurt ...", "OK then this is a crisis, let's do it". And then public health suddenly had to justify this new position, so you get a proliferation of "studies" along with ridiculous stories like "we were lying to save masks for healthcare workers" (think about the nonsensical logic of this for a second!)
> If you narrow it down to just studies that use the scientific method correctly and which are built on strong evidence, no convincing findings remain.
That's a significantly weaker claim than the "masks did nothing" claim you've made in other posts [0-1]. And yes, the Cochrane review stated an inconclusive result; it explicitly doesn't say that masks do nothing. Nor does it really support what you're saying.
Firstly, note that the Cochrane review reviewed the effectiveness of promoting mask-wearing. Not the effectiveness of mask-wearing. It heavily weighted studies with a low (40-50%) adherence to mask-wearing [2]. This limits what it tells us about how well masks work.
Secondly, the Cochrane review arguably overstates its case. There were two randomised controlled trials studying the effectiveness of mask-wearing during the pandemic. The first found that promoting mask-wearing in rural Bangladesh increased mask-wearing by about 30% and reduced the risk of COVID-19-like illness by about 10% [3]. The second found that promoting mask-wearing reduced coronavirus infection by about 18% [4], but the low sample size meant that it could only detect very large effects of 50% of more [5], and the adherence was again low at 46%. Many leading experts in epidemiology believe that this points to a small but worthwhile reduction in COVID transmission when mask-wearing is promoted [6], implying that correctly wearing masks hass a significant effect in reducing transmission.
Indoors. There were decades of research leading to the recommendations of mask wearing when symptomatic and only indoors.
All that fell by the wayside when mask wearing became a covid-time cult. A friend (with a degree in epidemiology) told me that if she tried to argue those points and doubts outdoor mask mandates she will be the immediately out of her job.
The covid-time environment of shutting down scientific discussions because policymakers decided that we had enough science to reach a conclusion should not be forgotten, it was a reasonable concern turned into a cult. My 2c.
This is the start of the end of generative AI music.
Without the entire music catalog, they are just not going to be able get the diversity and quality of outputs they once had, and the labels are always going to have their hands out wanting more.
Udio and Suno have demonstrated that okay-quality new music can be created by training on vast collections of pirated music, and, as expected, the music labels shut them down.