All of you would love the "obscure cities" series of comics (https://en.wikipedia.org/wiki/Les_Cit%C3%A9s_obscures), as well as some works by J. Abeille (https://en.wikipedia.org/wiki/Jacques_Abeille#Les_Jardins_st...)
183 karma · joined January 7, 2023
This is of course complicated further by the sheer pigheadedness of CEOs with a career half life of 3 years who come in, _revolutionize_ something by destroying it and antagonizing the workers, bloat HR a bit more and drive sales through something illegal; and finally move on, tallying that one a success.
A potential upside is: lowering the marginal cost of page filling drivel so much in conjunction with declining advertisement revenue, the whole industry collapses in on itself, taking opinion pieces with it. The baseline for getting out of journalism school would be "write better than an AI", and the reporter jobs themselves become high status / high difficulty again.
A potential downside is: nothing written ever matters again, all information is a rehashed AP cutting with prompt engineering, and AP / Reuters become a set of controlled bottlenecks.
The scary thing is: this will be decided by whether the paying customer wants to be informed, or feel the warm glow of confirmation bias?
Real talk now, I'm an "AI person" from big pharma, and we're quite up to date. topological neural networwks, diffusion models, QM neural potentials, large scale meta-learning, systematic active learning, we're doing it all. We also know that most of the time, a small bayesian GLM or random forest on run-of-the-mill descriptors actually works very well and fails predictably, which is important.
Data in pharma in particular tends to be sparse and shallow: a hundred datapoints clustered tightly in chemical space because that's what the process generates. Sharing the data can lead competitors to the precious IP you're protecting, which is why we also invest a lot in blind federated learning etc.
Anyway, the going's tough, but everyone is doing their best. nobody's dismissing AI at all... we're just more aware of the domain-specific pitfalls.
and the companies do not want to compete: most modern large companies only exist as a _vehicle for investment_, and as such compete for shareholders, not consumers. Due to this, short-termism dominates: average CEO stays 4 years at a given company, and performance of the company after they leave is largely uncorrelated with their carreer success, while share prices during tenure actually does.
So they'll suck the blood from us, crash everything, and get a raise when politicians ask them very nicely to try and fix it. And while share prices will go down on the S&P, billionaires will gte richer, and hedges will make bank...
- people are harping on bias, but I think they miss the point. Use the tool, get the tool result, and let god sort them out.
- Multilingual sources are core to the value proposition. Make it a community list with tags, e.g: a user from india add their favourite local UP political news source with tags [india] [politics]. then you can taylor the reports by interests (yes, like all news aggregation websites before. they do it because it's needed I think)
- likewise,store the summaries, then enable users to ask for a rewighing with their interests in the prompt, e.g: "I am a well educated, religious Emirati interested mainly in news about the wars and tensions in the middlke east, as well as car racing"
- get news by actual news, not by news source. E.g: get the embeddings of the news, cluster, get summarization of "best" sources of cluster with added info like likely primary source, histograms of reputation vs sentiment vs political leaning
- todays news give you also access to the comments, both on the site and e.g: on reddit and twitter. add maybe a comment summarization / sentiment analysis (e.g: 90% of comments on this new york post article were racist one-liners. Dicscusion on this Korean article revolve around effects on workforce mental health, etc)
- you mention cost: make accounts able to add their ChatGPT API keys, with an estimation of how much you'll use, ang keep these accounts free. For the others, warn that in time you'll add a small charge maybe?
Everyone I met who writes fortran (mainly for quantum chemistry, solid state physics programs) uses multiple compilers to ensure the highest quality code.