It's a huge win that we get perspectives other than the official narrative. And that it's triggering the people that are upset someone disagrees with what they're selling.
I disagree with the "misinformation" label generally. But I can see with voting or public health how there is a public interest in people receiving correct information (ignoring that authorities have lost all credibility on the larger anyway).
But when it comes to politics, or "science" if we insist on calling it that, an official narrative is counterproductive and everyone has a right to say what they want, even if it's "wrong" or you don't agree. TMZ may be a source of celebrity misinformation. Who cares.
It's my perception of this social justice stuff that it infects everything, because it's somehow become a magic wand that give people power over others. The whole point in this example is that resources from what is ostensibly an computer science research institute is working on fringe left political issues. It wouldn't be ok if they studied "intelligent design" or whatever the current far right is up to either, but politics has managed to infiltrate research so we're all forced to deal with cuture war stuff instead of working on computer science.
I'd say that's a pretty clear example of why woke==bad. The title reads like something that was made up by the Onion in an attempt to mock a charicature. This kind of garbage draws attention away from actual research.
Also, while it gets lost in the foundation model stuff, a major trend in computer vision is toward smaller, high quality datasets. Arguably CV had its V1 llm moment years ago with models trained on imagenet, which produced amazing general results but weren't good enough for much specific stuff.
If you look at what, e.g. Andrew Ng was talking about last year, there was a big emphasis on "small data" and getting good datasets.
That doesn't follow from anything. Research is part methodical slog, part lottery, maybe a pinch of intelligence. A few labs won the short term lottery here, and most researchers explored stuff that didn't get headlines. (And to be fair, OpenAI built a great product that catapulted lab research into popular view).
There might be some argument on other metrics - publications, students trained, lectures, recognition, whatever, that show this institute is lagging. But not being part or llms implies nothing about their success or failure.
Not focusing on LLMs isn't a major sin imo, if I was donating money to an institute I'd rather they were doing something unique than churning out another also-ran llm. Researchers have diverse interests and expertise, and the field has way more depth than the current thing, it's not obviously bad they were working on other stuff.
That said, if the examples he gives of what they were working on are representative, it implies they are spending their time chasing trends instead of doing fundamental research and chased the wrong ones. I'd suspect they re doing more than what was implied though.
This may be a "don't hate the player hate the game" type situation. Influencers or aspiring influencers are doing what they do because they are working inside a construct that's set up to reward them for it. They are not blameless, but they are not conspiring to create that damage that social media had done to society, they are in s sense being put up to it by the platforms and users (who by the way I don't think should get a free pass either - PSA: delete twitter)
Most construction relies on fit young men to do the "heavy lifting". It would be interesting to see how the dynamics changesd if technology moved the needle on the brawn->brain scale, so that the core value prop of a construction worker was not their fitness. I'm not convinced it would be for the better.
Also just want to add that working in a physical job for 50 years is much better for your body than sitting at a desk for 50 years.
I have never laid brick but I've done a lot of shingled roofs. The "fun" part of shingling is long stretches of full sheets in the middle of the roof. This goes quickly and is rewarding. The less fun, and what occupies the time, is where there is a chimney or a valley or a dormer or some other feature that you need to figure out and cut some flashing for, and custom fit each shingle that butts up against the feature, seal with tar sometimes, etc.
I picture a lot of automated solutions focusing on the "in the field" part where the work is easy and rewarding, and glossing over the manual intervention and craftsmanship that is needed for all the nonstandard bits. I don't know about bricklaying though I imagine it has similar characteristics, where a square windowless wall goes very fast, and could be pleasant work, and all the time is spent dealing with windows and the frame not being level and other edge cases.
Yes agreed, sorry I was thinking in terms of a new product, testing the value prop with users, which is probably better stated as requirements gathering.
I used to work in a family construction business and now I do machine learning contracting. When I compare the value you get for 300k worth of construction to what 300k of ML specialist time buys you, it's hard to agree with your comment. I understand the reasons, but generally I'm blown away by how cheap most labor intensive things are, compared to how much people will pay for software projects.
What you're saying resonates for so many new products. I find that most time or labor saving ideas save time on something that is not actually the problem, or address a charicature of the actual workflow. I dont think it means we should give up and stop trying, rather that its critical to test with real users IMO before trying to sign deals on a product that doesn't make sense.
I have no idea how the spectrum of political views can be divided into two discrete categories. But a plausible reason one could be more subject to "misperceptions" (hopefully people see how scary it is to have that be the way that a political view is characterized) is that some people are more free thinking than others and don't just accept orthodoxy as fact.
I'm not trying to say that my alternative hypothesis would follow some kind of party line, just hint at how there could be many interpretations of this study, if you don't already reject its entire premise of trying to bucket people and then find intellectual differences between then.