Godfrey642

Just interested in peoples attitudes to AI
Please boost

#genai #ai

love it love it love it
0.6%
makes everything in my life easier
0.9%
the speed its being developed is worrying
9.3%
needs regulating in public hands
16.9%
I wasnt asked if I wanted it
20.2%
its dumbing down the world
24.3%
huge energy cost is adding to climate breakdown
27.8%
Poll ended at .
Dec 29 at 12:27pmWeb
Show threadPatrick Loftus 🖖Dec 29
@Godfrey642 I’m interested in the tool and RAG abilities of small and specialized locally runnable models within low power constraints.
Show threadGodfrey642Dec 29
@pwloftus
Interesting.
Please explain.
Show threadGodfrey642Dec 29

@pwloftus

Im disturbed primarily about the energy costs.

Also the AI that is impersonating humans in rogue videos without permission. Truth is being compromised yet further

Show threadPatrick Loftus 🖖Dec 30

@Godfrey642 I run small models locally on 3 laptops.

1. Nvidia 1650m with 4GB of ram, max of 30 watts.
2. Nvidia 1070m with 8GB of ram, max of 125 watts.
3. Nvidia 4090m with 16GB of ram, max of 175 watts.

The 2nd laptop also runs my mastodon instance.

The LLMs do a fairly good job of parsing natural language to intent. For example a math specific tool allows for taking a user’s natural language prompt, parsing it, then passing it to math functions that return accurate and truthful answers rather than inferred answers from the model.

Training of models is absolutely energy intensive. Running models, not necessarily.

The AIs /LLMs don’t run without a prompt, human beings are ultimately behind lies and deception and are also responsible for the outputs they have prompted for.

I agree the loss of truth and impersonations are very valid concerns.

Show threadTim Ward ⭐🇪🇺🔶 #FBPEDec 30

@Godfrey642 Depends what you mean by "AI", doesn't it.

Systems that aid humans with pattern recognition to detect fraud and cancer, as we've been using successfully for decades, good.

The newer systems that generate garbage text / audio / video, bad.

Show threadTim Ward ⭐🇪🇺🔶 #FBPEDec 30
@Godfrey642 In particular, if there were no AI fraud detection in the payment process you'd find it a lot less easy and/or safe to buy things.
Show threadGodfrey642Dec 30

@TimWardCam

I personally 100% agree.
So is the only remedy regulation? I know Americans bash on about Freedoms to do and say whatever... but outside of USA we talk about freedoms FROM malevolent practices.

Companies - particularly tech companies - are huger in value than entire countries now. They are unelected. Often corrupt in practices.

What safety nets (Freedom) do we have against (from) unfettered, profit-centric, unscrupulous boards of directors and shareholders?

Show threadHurricane DaveosaurusJan 1
@TimWardCam Even in those systems the bad may outweigh the good. Google "Robodebt" for more info.
Show threadZ_Zed_ZedDec 31
@Godfrey642 "It's not AI, it's marketing" wasn't an option.
Show threadMaxilo OveritDec 31
@Godfrey642 It needs to stay in the lab, in air gapped PCs, for at least the next 20 years. No generative ai should be publicly allowed at all, let alone employed or given access to the web.
Show threadGodfrey642Dec 31

@aka_quant_noir

Do you know something the rest of us don't?
Or is it just general hatred ?

My concern is that is in the hands of unelected oligarchs who are clearly not concerned about us, the people. They have more wealth than entire nations.

I see AI in the same light as Biohazards Level 4.

Show threadMaxilo OveritDec 31

@Godfrey642 I have no hate in my heart. But caution, extreme caution, regulation even, is warranted before any dangerous tech is brought to market, or allowed access to our minds and stolen data.

Should have been. Pandora.

Show threadGodfrey642Dec 31

@aka_quant_noir

Agree.
Why do you say "regulation *even*..."

Show threadMaxilo OveritJan 1

@Godfrey642

Because regulation is a dirty concept for the rich klepts running capitalism.

Show threadGodfrey642Jan 1

@aka_quant_noir

The rich klepts need to be put back in their boxes. Its not about them. It's about what the majority want.

Americans need to focus less on freedom TO do whatever they want, and focus more on freedom FROM the evil of corruption, domination, manipulation, exploration. Regulation is about that latter freedom.

Show threadReneDamkotDec 31
@Godfrey642 the option ‘I hate it with a vengeance’ is missing. @adipoeserPursch
Show threadsiderealDec 31

@Godfrey642 Personally I still have yet to see a use case that is in any way... useful... or could make me more money than it costs to use. Until one of those happens, I'll keep wondering what the point is.

The text or code it generates needs to be edited and that often takes longer than just doing it right in the first place. The images it generates are tasteless and boring. The videos it generates seem to mostly be harmful propaganda and/or deepfakes.

More people are realizing this daily.

Show threadViolet MadderDec 31

@sidereal @Godfrey642

The point is, it's a weapon. That's THE point, as it is currently being designed and deployed.

Show threadBoneHouseWasps🔶🇬🇧🇪🇺Jan 1
@violetmadder @sidereal @Godfrey642 "AI is a weapon" is precisely the right framing. And will remain so until its use is properly regulated.
Show threadPaulNickson 🕊️🎛️🎼Jan 2
@violetmadder @sidereal @Godfrey642 And I think that is the intention.
Show threadAmber  (aka kemona_halftau)Dec 31
@Godfrey642 generative "ai" specifically or just machine learning in general?

- 💜 amethyst
Show threadGodfrey642Dec 31

@kemona_halftau

Either
Both

Show threadHurricane DaveosaurusJan 1
@Godfrey642 Not included was my preferred option: Kill it with fire.
Show threadGodfrey642Jan 1
@Daveosaurus
Or a silver bullet and a stake through its artificial heart.