Recent searches
Search options
In our little elite tech bubble, we can have lots of opinions. But I'm finding it interesting that how #gAI is being used is now, apparently, showing up as consumer backlash
https://www.cnn.com/2024/08/10/business/brands-avoid-term-customers/index.html
I had a guy cold call, in person, at my job. Met him in the lobby - he had donuts - and wanted me to hear his elevator pitch for an IT product.
OK.
It was for a network device that had built-in 'AI' for some reason or another, and I stopped him right there, "Listen, my eyes glaze over when I hear AI. It's a buzzword, and it's not going to sell me your product..." He went on a bit, and handed me his card and the donuts. I have no idea who he was or his company. Card into the bin.
@boelder First, that's weird, second, I completely agree that magically sprinkling "AI" on your hardware is a warning sign. We've been told for months that VCs were throwing money and any company with AI in their pitch. I guess this is the inevitable result
@boelder @scottjenson “AI” is being sprinkled as a sauce on top of things that may fit some generic description of “AI” - but has been used for decades.
It bugs me. Suddenly things that were just “algorithms” are now “AI”.
I feel that machine learning and neural nets are important technologies that is having some real world uses now, and a lot more soon. Even LLMs have productive uses.
But wrapping absolutely everything in “AI” is just hype.
@ehurtley @boelder @scottjenson
The #SparkleSickness 
is going to be endemic: https://www.superversive.co/blog/sparklesickness
@ehurtley what productive uses have you come across for LLMs? cause so far i haven't seen any @boelder @scottjenson
@mensrea @ehurtley @boelder There are a LOT. Please keep in mind I have lots of issues with both the energy and ethical issues with present LLMs (and I hope they will eventually be fixed) But I've found them very useful for programming, especially small little throw away prototypes. I've also found them helpful for creative writing, not the final output but for creating a range of alternatives for me to pick from.
@scottjenson @mensrea @ehurtley @boelder
What sort of alternatives do LLMs give you in your creative writing process - and not to be too blunt but can you call that creative if it is being generated by a computer?
@drmambobob @mensrea @ehurtley @boelder not to be too blunt but have you tried it? I think you're romanticizing the creative process. There are lots of very little things that llms can help with that don't remove the human from the creative loop.
I'm still skeptical mind you. Just saying you're being a bit strong.
@scottjenson @mensrea @ehurtley @boelder No, I haven’t tried it, and I’m not interested. I’d rather spend time struggling through my own variations to refine my ideas. I find the whole process helps me organise my thoughts and present arguments that are coherent and consistent with what’s in my head. It’s not romanticising, I find externally generated starting points and frameworks restricting.
@drmambobob @mensrea @ehurtley @boelder You do realize you could use that same argument to never use laptops and mobile phones right? Using a pen and paper actually is far better creatively than using any digital device. But that doesn't mean (and I assume this applies to you as well) that there aren't uses for computers?
@scottjenson @mensrea @ehurtley @boelder I disagree. Those things don’t generate words that are linked with thoughts.
@drmambobob @mensrea @ehurtley @boelder Look, I don't want to be snarky or disrespectful here. There are lots of reasons to be skeptical, no #generativeAI fanboi-ism here.
However, I am saying that throwing EVERYTHING out just because some stupid people have overhyped this is short sighted. The sky is already falling on #LLMs, the bubble is bursting, but there will be very powerful uses of this tech.
I'm reminded of this quote from Arthur C Clark:
@scottjenson @mensrea @ehurtley @boelder I’m not throwing everything out. There’s a big difference between ai-assisted tools like grammar suggestions, spell checking, calendar suggestions, or photography tools, and generative ai. Even with the latter I’m not talking about generating text for mindless menial tasks like standard email replies or filling out forms, but about creative writing.
@drmambobob Well, in my case, I used LLMs to help my writing quite a bit. I research everything from quotes, historical notes, books, and trends. But it's the ability to ask it incredibly vague questions and get surprisingly helpful replies that I found most useful. I use it like a thesaurus/encyclopedia with an astoundingly helpful index. I don't think it's "killing my creativity" in any way. In fact, it's magnifying it. Of course I validate anything it says, that's fairly trivial to do.
@mensrea @ehurtley @scottjenson
I actually have used ChatGPT a couple of times to write bash scripts to do some mundane tasks... Nothing elaborate, but it was pretty cool.
@boelder @scottjenson more importantly, how where the donuts?
I tossed them in the breakroom... They were from a local donut company that's really good, though.
@scottjenson I wonder to what extent this is related to people’s deep intuitions about enshittification (which seems very widespread at this point even if very few people know the word).
@luis_in_brief Yeah, there are enough move-fast-and-break things stories about #gAI and even consumers have got the memo.
To be fair, I do think it's possible gAI can be used in interesting (and ethical) ways. It's just a challenge for a lot of companies that are using it solely to "save money"
@scottjenson right, or using it without QA and so not understanding the rough edges that consumers are seeing either directly or in the media.
(I use it productively myself in some ways, but where I control it, understand shortcomings, etc etc)
@luis_in_brief @scottjenson
In this April 2024 podcast convo with Ezra Klein, Nilay Patel said that, for kids, “AI has become shorthand for shitty”.
The kids •are• alright: https://podcasts.apple.com/us/podcast/the-ezra-klein-show/id1548604447?i=1000651522107
@paninid @luis_in_brief @scottjenson the kids are right
@Jennifer @paninid @scottjenson also the kids are intuiting something important about the now-broken relationship between capital and innovation.
@scottjenson @Viss tbh the best part about the whole thing is that there's a Taco Bell Distinguished Professor of hospitality business management. I wonder what the chairs of other endowments think about that.
@darkuncle @scottjenson @Viss why did I have to get all the way to this comment before someone noted this professor is the Taco Bell Distingiished Professor?
@scottley @darkuncle @scottjenson @Viss I am a Blimpie valedictorian and I will not allow my fast food academia peers be slandered like this
@samhenrigold @scottley @scottjenson @Viss I lol’d
(to be fair, I ate more Taco Bell as an undergrad than probably any other food source - and lived to tell the tale!)
@darkuncle @samhenrigold @scottley @scottjenson @Viss You should call them, maybe a honorary doctorate Dr T.B. is in the cards?
@darkuncle @scottjenson weirdest timeline
It's how they were prepared to win the Franchise Wars and be the only restaurant in 2032.
@scottjenson @jhulten @darkuncle
Nevertheless, here's the scene for everyone's enjoyment: https://m.youtube.com/watch?v=AlcDHlK_RoY
@karttu @jhulten @darkuncle genuinely honest. Thank you
@scottjenson @karttu @jhulten !!! in that case I am super excited to share with you: https://www.youtube.com/watch?v=s-LzYwbYkQs
@scottjenson @karttu @jhulten (great film from the 90s, Dennis Leary in particular was excellent)
@jhulten @scottjenson totally underrated 90s movie there, well played
@darkuncle @scottjenson @Viss I know for a fact that the Denny's Stacks of Pancake Professor at the IHOP Institute for applied statics really looks down on him as "not real science"
@darkuncle @scottjenson @Viss
Partner just now: "Yes, I'm the Taco Bell Chair of Hospitality. ... oh, you're with Arby's? That's nice."
@unikitty @darkuncle @scottjenson i'd like to be the 'honorable chair for the committee of putting weird nondescript 'protien' and sawdust in troughs and laughing all the way to the bank when the FDA allows us to call that food'
@Viss @darkuncle @scottjenson
One of those times I'm glad to keep a vegan diet lol. Their bean slop can't be as gross as their "meat" slop.
Taco Bell Distinguished Professor‽
@scottjenson Had to buy a new refrigerator last year (in a rush, inevitably). It has "Artificial Intelligence" plastered on the front despite limited other "smart" features.
Turns out they want you to install their app on your phone which tracks your location so it can "learn your routine" and increase cooling when it thinks you are returning from the grocery store. Privacy nightmare 
“One of the reasons why people are not willing to use AI devices or technologies is fear of the unknown,”
But, just before that the article says that consumers didn't like google AI results because they were sometimes incorrect ... making it unreliable.
That's not "fear of the unknown" it's not liking something that doesn't work. This is my own issue with AI... the results just aren't that great.
@futurebird @scottjenson
I had a few tries with simple questions, it came back with some sort of Reply Guy answers..
A good movie, a good book, a good painting: Sometimes somebody gives a hint, or you discover the hidden clue, almost another dimension , by yourself.
"Between The Lines".
I don't think any AI robot can see that other dimension.
Or create something like it.
Just mimicking .
@futurebird@sauropods.win @scottjenson@social.coop yeah I think the most damning thing I can say about AI right now is that it's useless
I mean it literally does not help, every time I have used it I get slowed down, it's kind of just impossible to seriously consider AI as anything more than a toy
@futurebird @scottjenson It's fear of the answer being unknown TO THE AI.
@apophis @futurebird @scottjenson Yes, quite.
Except.
I would say EVERY answer is hallucinated slop. It's just that sometimes what it hallucinates happens to be the truth, or close to it.
@futurebird @scottjenson It's also extremely compute-inefficient. It's the most expensive way to get something done half as well as a professional could do it, and its inaccuray means you still need a professional.
@scottjenson I'm convinced this isn't the straight-up Ludditism it appears to be. Consumers don't hate AI as such. But they've learned that in practice its use in consumer goods is a reliable marker for extreme hostile intent. AI isn't being used to help consumers; it's being used for spyware, price fixing, redlining, fake customer service that doesn't actually help customers, and other forms of legalized crime. Often there isn't even any substantive use of the tech; it's just "AI-washing".
Google used to sometimes give random quotes at the top of search results, with links and everything. Often, the quotes answered a different question than were asked, or implied the opposite from what the full text actually said.
Now, I get "ai summaries", and it's harder at a glance to verify what the original text said, b/c it's concealed. So I always skip over it to the real results.
@scottjenson WTF is a "Taco Bell Distinguished Professor"?
@scottjenson My washing machine has Wifi and "AI" programs. The product page says "it personalizes washing by remembering your habits, suggesting cycles and displaying timely information", with a footnote leading out to "Server-based big data analysis is not available in CIS, China and in markets where the server is not available."
I do not understand how AI or big data is improving anything here; as the thing is lacking a feedback mechanism on whether or not it did a good job, it cannot learn anything about its dosing or cycles or whatever to optimize things for me. Or do they mean that the "AI" is just a frequency table so they can show me the option I use most often at the top of the program list? That'd be massively underwhelming.
The "big data" can however be used to profile me and can easily be combined with other IoT "AI" devices, it increases the attack surface on my home network and I can be practically sure that whatever "big data" is collected will be leaked within the next 5 years.
@scottjenson On a related note, any network-enabled AI feature also opens the door to network-enabled subscription services - as BMW and, surely, many other auto makers will introduce in the future. Whenever it's time for me to buy a car, I'll pay a premium to ensure it won't be able to do OTA updates (and another premium for physical buttons instead of cheap touchscreens)
i eventually got over it but by then we were already used to going somewhere else
The only new feature I can think of wanting is a Fawkes image cloaker (or similiar) built in to my phone.
I'd pay for that