this post was submitted on 04 Oct 2023
84 points (92.0% liked)

Technology

59665 readers
3556 users here now

This is a most excellent place for technology news and articles.

Our Rules

  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots

founded 1 year ago
MODERATORS
L3s
[email protected]
[email protected]
L4s
enu
84
Why Big Tech's bet on AI assistants is so risky (www.technologyreview.com)
submitted 1 year ago by [email protected] to c/technology
18 comments fedilink hide all child comments
top 18 comments
sorted by: hot top controversial new old
[–] bill_1992 47 points 1 year ago (2 children)

This whole thing is basically a nonstory when you realize how much money is in tech. Meta changed their name and sank billions on an idea that everyone thought was stupid from the beginning, and they're still fine.

Putting a billion into the flavor-of-the-month that has like 10% chance to be the next big thing is a no-brainer when you're printing multiple billions in profit doing nothing, and have a lot more cash on hand.

The real story, is how wealth inequality and monopolies have essentially allowed the rich to waste tons of money chasing more wealth while having almost no incentive to provide value to society. Who gives a fuck about hallucination and prompt injection? It's all trivial details that VCs are giving away billions to eventually solve.

[–] [email protected] 4 points 1 year ago* (last edited 1 year ago)

This parallels the height of GM, when they put money in everything from satellites to ATMs. Sure, at the time there was plenty of money to dump on such ventures. But eventually, those bad bets caught up to them. It seemed like a "no big deal" when a hugely profitable company wasted billions of dollars on bad investments. But each one of those bad investments represented a lost opportunity for a good investment. Eventually, the cash cow at GM ended, and the company is left with nothing but huge debts and worthless investments. Any tech company that is just buying the equivalence of lottery tickets is probably destined for failure too, no matter how profitable it is right now.

[–] pdxfed 1 points 1 year ago* (last edited 1 year ago)

Well put with that relevant user name for the Dem who cemented wealth inequality with Nafta that Reagan had worked so hard to get the ball rolling on.

[–] AA5B 20 points 1 year ago (2 children)

Amazon is losing money on basic voice assistant, but maybe if they make it more expensive , it will become profitable ?

[–] [email protected] 10 points 1 year ago

I mean, that’s a little like asking, “How many swift kicks to my nuts before it makes me a billionaire?”

Because it’s one of those cost evaluation situations where they thought it was a shot in the dark at first, but by now it’s clearly a loss. So, the whole thing feels a little like, The Producer’s, something isn’t smelling right for any outsiders.

[–] Fez 3 points 1 year ago (1 children)

Alexa is trash and its entire implementation is trash. I’ve tried it with a smart home and it’s a nightmare.

[–] [email protected] 6 points 1 year ago* (last edited 1 year ago) (1 children)

It's...okay. What is garbage is that after billions of dollars spent, it has barely improved in the last 5 years. In some cases gotten worse.

And don't get me started on Siri.

However, if Alexa/Siri had integrated with ChatGPT instead of Wolfram Alpha and "Amazon contributors" I might feel somewhat different. The info would definitely not be less reliable.

[–] Fez 3 points 1 year ago (1 children)

I switched from Android to iPhone a few years ago. I still miss Google Assistant so much. 😬

[–] [email protected] 5 points 1 year ago (1 children)

The part that kills me is that Siri was (effectively) first. They had probably 2 years' lead on everyone else. Somehow they squandered that - pretty quickly - fell behind, and in some ways got WORSE. And it's been 12 years.

[–] AA5B 1 points 1 year ago* (last edited 1 year ago) (1 children)

The one thing Siri has above all else is being local where possible. Personally I thought one of the biggest announcements from last month’s Apple event was that the new watch is powerful enough to do voice processing locally.

Of course that also may be one of the reasons Siri lags the cloud voice assistants

[–] [email protected] 1 points 1 year ago

That does sound good.

It would be nice for all these things to be a lot more local. At least local control things should be able to be done without a cloud service.

[–] BertramDitore 19 points 1 year ago (1 children)

Because people like me genuinely don’t want this crap?

[–] [email protected] 5 points 1 year ago (1 children)

Like never? Even as it gets better?

[–] BertramDitore 4 points 1 year ago

For me, that depends on a whole lot of things. Who owns it? Who pays for its operation? How much and what access does it need? What verifiable privacy protections are there? How transparent are its processes?

I’m just generally suspicious of companies whose products are ultimately meant to be integral to our daily lives. These kinds of tools will undoubtedly need access to a huge chunk of our personal and professional data to work effectively. I’d rather not interact with the world through the lens of someone’s corporate vision. With the right protections? Sure, I could see some cool and creative use cases. But that unfortunately brings me back to my general skepticism.

[–] [email protected] 17 points 1 year ago (1 children)

Me looking up the exact order of commands I have to give the machine to get connected to a real first level support employee.

[–] [email protected] 4 points 1 year ago

Me, an "AI written" regurgitated article on a content mill website uncaringly misleading you about the best way to get connected to a purple banana fridge.

[–] [email protected] 10 points 1 year ago

Gonna just buck the trend and say that this AI push has me excited for the future. It's easy to be a nay-sayer, but I genuinely believe the leaps made in AI in just the last year are amazing.

The author clearly doesn't like AI, and completely mischaracterizes Mistral AI for things their models could say, but doesn't consider at all why unaligned models are useful in developing your own.

The author likes to highlight that sometimes an AI will make things up, a phenomenon known as hallucinating. Hallucinations could also be called "creativity" in certain contexts. This isn't always a fault, especially when creativity is the intended purpose.

The author pointed out how it's possible to prompt engineer out sensitive data, and how there's a lack of privacy... which isn't a problem with the tech, but rather tech companies.

The technology used behind the scenes with ChatGPT isn't exclusively for text generation. I'm seeing it appear in speech to text / text to speech applications. It's showing up in image and video editing. It's showing up in ... well ... images/movies of an adult nature.

You're probably already consuming AI generated content without even realizing it.

[–] Usernameblankface 4 points 1 year ago

Hmm, an unfinished technology has an exploit no one building it expected. Heh, welcome to the Internet.