hobom

Like the OP, I also greatly respect the work Rachel and you did for AI.

Nevertheless, I can't help but think that you are seeing this issue to negatively.

"The point of the article is that AI as a research tool is insufficient to result in improvements to patient outcomes on its own." This seems unlikely if you consider this question as-is. Past technological improvements has made healthcare better overall without requiring societal changes necessarily. Take mRNA vaccines, a technology that has improved Covid-19 outcomes tremendously. Sure, certain groups have less access than other groups, but surely even marginalized groups are better off overall because of the existence of these vaccines. Health is not a zero-sum game.

And I think this negative attitude also misses the potential of AI by default. Yeah it sucks that not everyone gets MRI access, but those that do will benefit, including marginalized groups. I guess it feels wrong to some people to express a positive sentiment at an unjust state of affairs, but improved diagnosis and treatments translate into lives saved.

You also have to compare AI to the status quo. Sure, AI will have biases, but so do humans (as Rachel points out!) and the decisive question is whether AI has less biases (similarly to how safety of self driving cars is judged). With AI you at least have a chance of analyzing the decision making, and making it objective. We should be extremely excited about this possibility!

It's basically all present in the article

While true in general, it is really not that necessary to have a representative sample for checking whether your software works. With large enough sample size a convenience sample will find almost all of the bugs that telemetry would have found.

Comparing against 1990 is wrong though, as we should compare against the start of widespread social media use (2010 at least), and it's also pretty likely that other OECD countries are lagging behind with widespread adoption by a few years. If we don't see an increase from 2015 to 2022 THAT would be counterevidence.

This is not an accurate reading of the Conclusion section. He just sums up what has been observed and does not mention any causal factors. Earlier, he explicitly says that causal factos will be covered by other posts. While he indeed links to another article, he claims at no point in this article that coddling is a factor here.

Donald Trump never attended an MBA program.

That is not true. All food in the supermarket is taxed the same.

The just-world fallacy is strongly in play with many comments here. Because those foundations and charities are massively affected by this scandal, there MUST be something they did wrong, right? Surely they should have vetted their donors better? And when making these arguments, we totally forget that people specialising in vetting people, like investment funds, journalists and so on, completely fell for the guy. If you want charities to implement a standard that would have prevented them from taking money from SBF, congratulations, you now have a process that prevents them from taking money from basically any billionaire.

Calling Israel an appartheid state, and your specific examples betray a state of mind which applies arbitrary standards to Israel that never get applied to literally ANY other country in the world.

Some of your specific examples: "Illegal Settlers living in Palestine can vote in Israel but not Palestinians." Americans living abroad in France can vote in American elections, but note the French people. Therefore the US is an appartheid state.

How is controlling your state border via a wall appartheid?

I have always wondered whether inducing exhaustion through exercise and strainous activities throughout the day could help overcome insomnia.

Unfortunate consequence of a legacy legal system. Uber drivers are clearly neither quite like employees nor are they entrepreneurs, and the law should come up with a fitting category that ensures they are protected from exploitation but continue to enjoy some of the freedoms associated with the gig economy.

His science journalist part is also not convincing. Neil deGrasse Tyson far outstrips the popularity of science journalists, because unlike other scientists he is not boring to everyday people

I think there was some reddit data hoarder initiative that did that a few months ago.

I think your tweet is in bad faith. Whatever the CEO thinks of Amazon warehouse workers' salary and working conditions has no bearing on the quality and outcomes of the lambda school curriculum. This Twitter style argument of "Look at this person having awful opinions" really does not contribute to anything.

If you find lambda school to be predatory, then it's better to just provide evidence for that.

Thank you, this is just what I was looking for. I like reading non fiction, but have found it incredibly hard to determine whether the contents are actually correct, since most reviews are not done by subject matter experts. I am tired of reading and enjoying psych books only to find out that 90 percent of it is misleading or untrue.

There have been countless Googlers who came out to support Gebru. Did Google fire all of them? Or maybe Margaret's support was particularly outspoken, but I doubt it.

I am kind of jaded, but despite the interesting approach, I predict that this research will not end up being deployed in time to make any difference. The real world pipeline just takes too long.

This includes some true things about data collection issues, but I cannot agree with the main thesis that ML algorithms are about power. If anything, they shift power to the patients because now the decisions can be checked and questioned at many levels. An algorithm will not send you away because it is to tired too take your complaints seriously.

So, using algorithmic decision making should be as bias free as possible, but there is no way that across the board they will be more biased than humans are now. If people care about marginalized communities, they should push with everything they got for, not oppose, ML decision making.

The AstraZeneca trial was the one to use the meningitis vaccine.

Thanks a lot for posting. Are you involved with the project? If so, great work, and I am excited that you made transfer learning work for this task!

I am interested if you have given the following suggestions some thought: 1) The number of false positives is quite high, which is a challenge for a screening tool. However, the app could be used as a precursor to a quick antigen test: Rapidtests have basically no false positives, and your model basically has no false negatives. 2) The model could in principle be extended to be able to detect that a user's cough has changed. The user would record his "healthy" coughs as a baseline, and the model would try to detect a change when the user gets Covid-19. In principle this approach has the advantage that it is a intra-subject comparison, so the model could deal better with the particularities of a particular user (their mic, their voice and so on). Of course gathering the required training data might be a bit harder, since you now need two recordings from each user/patient.

The paper mentions that this model is currently being tested in several hospitals. How quickly do you think it could be rolled out?