I asked ChatGPT, Gemini, Grok and DeepSeek to tell me about a contemporary Scottish indie band that hasn’t had a lot of press coverage. ChatGPT, Gemini and Grok all gave good answers based on the small amount of press coverage they have had.
DeepSeek however hallucinated a completely fictional band from 30 years ago, right down to album names, a hard luck story about how they’d been shafted by the industry (and by whom), made up names of the members and even their supposed subsequent collaborations with contemporary pop artists.
I asked if it was telling the truth or making it up and it doubled down quite aggressively on claiming it was telling the truth. The whole thing was very detailed and convincing yet complete and utter bollocks.
I understand the difference in the cost/parameters etc. but it was miles behind the other 3, in fact it wasn’t just behind it was hurtling in the opposite direction, while being incredibly plausible.
This is by no means unique to DeepSeek, and that it happened with specifically DeepSeek seems to be luck of the draw for you (in this case it's entirely possible the band's limited press coverage was not in DeepSeek's training data). You can easily run into it from trying to use ChatGPT as a Google search too. A couple of weeks ago I posed the question "Do any esoteric programming languages with X and Y traits exist?" and it generated three fictional languages while asserting they were real. Further prompting led it to generate great detail about their various features and tradeoffs, as well as making up the people responsible for creating the language and other aspects of the fictional languages' history.
The problem with the App Store model is that the app could just be switched off by the powers that be. It would be better if something like this could be built into the OS. If one decentralised use case took off, then there could be other applications, like hosting the internet archive, wikipedia or LLMs, or digital cash. Might need waystations to get into rural areas but it sounds like the best long term way to secure the free internet.
Google currently has an advertising campaign for Gemini (in conjunction with Netflix!) which is all about how you can use AI to tell you what the key episodes are so that you don’t need to watch the whole thing. If that isn’t an admission that most of it is filler I don’t know what is…
Seems like autonomous personal air travel would be a far, far simpler problem to solve than road-based driving? Far less edge cases, once you are in the air the only obstacles you really need to worry about are other vehicles, birds, leaves and the odd stray plastic bag. With VTOL the whole thing could be an order of magnitude easier than driving. By the same principle, sea travel would also have less edge cases, at least in calm conditions. Is it the case that people aren’t seeing markets for those applications of autonomy or is there some other reason why there isn’t the same hype in those areas?
Vehicle cost; fuel cost; takeoff and landing; handling of edge cases (there's a reason that commercial airlines still have pilots despite autopilot being very effective in 90% of conditions).
A: Air travel generally is a far simpler problem to solve, because there's national standardization of the air navigation system, the airport paint and signage, and very clear rules of separation with very few edge cases. And yet we do not have anything close to self-flying planes no matter how much money you throw at it. And we're not even close to getting there. Whereas with cars, none of that is true, municipalities get wedged into violating their own equivalent of human interface guidelines all the time. Worn paint and signs for years, busted cross walk signals, confusing intersections, human driven cars that consistently do not follow the rules but in inconsistent ways.
B. Personal air travel has a significant regulatory burden in the transition from ground to air. The ground is city + state regulated. And immediately once airborne it's FAA regulated, but not under any kind of Air Traffic Control, as almost all airspace below 1200' above ground is uncontrolled. So ATC has nothing to say about it, and no central mechanism for negotiating conflicts.
Where are you allowed to takeoff and land is easily figured out today: airports only. And that's because it takes all kinds of things into account like obstruction clearance, noise abatement, anticipating engine failures and crashes. We can't even automate commercial flights - it might seem like it's more correct to say we don't automate commercial flights, but we could. That's not really true. The amount of changes to the air traffic control system, and on-board equipment for airplanes, is presently so cost prohibitive that it is effectively a "cannot be done".
Facebook jumped the shark when they started trying to encourage private individuals to “create engagement” and pay to boost their posts. If internal marketing jargon is leaking into your UX, you clearly have serious problems with how your staff are being incentivised.
Facebook could have become an engine that allowed society to maximise human capital and but they sacrificed all that potential at the altar of growth hacking and bleeding advertisers dry.
I remember when that first came out, and was based on how many friends you had. It would have cost me $90 to promote my vacation photos to a group of people semi-randomly selected by Facebook and chance from my friends list, and for people with more ‘friends’, they wanted up to $500, if I recall.
Socially, it is extremely out of touch to even offer such a service to individuals.
Nuclear weapons are the reason why the competing global powers have found new ways to compete and project power abroad. Global wars have mostly moved from being violence-based to being fought via propaganda and economics. It probably mirrors the way smaller communities moved from physical to intellectual competition.
Why are we still teaching assemby and C over development languages more popular, enjoyable, and accessible? It is not a good career path, it is not popular with kids or with the masses, and it is a lot more expensive than any other genre of software development.
That is comparing apples to oranges. C and assembly are what most applications and operating systems are built on in one way or another. Classical music is not the root of popular music nor does it carry over to other music genres that much at all.
Also, the way that music is taught in band rooms, is based almost entirely on sheet music and nothing on actually creating music. In fact most of the people I know who were very good classical band players had very underdeveloped ears and sucked at music creation. If you program in C, you will become a better python/js/etc. programmer, the same can not be said for classical music.
Plus, most computer science programs have switched from starting with C to starting kids off on Java or Python, so that argument does not make any sense. Most public schools do not even teach anything other than jazz or classical music.
Music is frequently taught at schools but not pushed as a career choice, so I'm not sure that the original complaint holds (at least not everywhere). Almost all children where I grew up studied music in some form at school and virtually none took it as a career, those who did were amply warned about the risks. This was considered unremarkable.
DeepSeek however hallucinated a completely fictional band from 30 years ago, right down to album names, a hard luck story about how they’d been shafted by the industry (and by whom), made up names of the members and even their supposed subsequent collaborations with contemporary pop artists.
I asked if it was telling the truth or making it up and it doubled down quite aggressively on claiming it was telling the truth. The whole thing was very detailed and convincing yet complete and utter bollocks.
I understand the difference in the cost/parameters etc. but it was miles behind the other 3, in fact it wasn’t just behind it was hurtling in the opposite direction, while being incredibly plausible.