How I Learned To Stop Worrying And Love AI

It's your future - do with it what you will.
Join Team FranLab!!!! Become a patron and help support my YouTube Channel on Patreon:
#franlab #rant #AI
- Music by Fran Blanche -

[davidliddelow]

I don’t think ai is ever going to wake up terminator style and suddenly control the world. Its more likely we will make an ai that does everything for us, put it in control of everything and the subsequent laziness is what will kill us.

[imqqmi]

"Who's crazy now?" Absolutely brilliant!

[pibyte]

"It's your future! Ruin it!"

Fran Blanche, 2024

[adenshures]

Get rid of all the smart phones. That's a dream come true.

[skippytheaustralian]

00:01:49:

Did You say "Colossus"?
I sing Daisy ...

[rodc]

So essentially AI could become a HAL but without the shutdown button - a chilling observation and a reality check the entire world needs to hear.

[KatharineOsborne]

I agree with you over the long term (with long term being in the next 10 years), but I don't think it's possible for LLMs to become sentient/conscious in their current form. The best descriptive I've heard for LLMs is that they are crystalised knowledge, because they have a training run (the crystalisation) and then they are used in a static form. Yes refinement and retraining happens, but not on a model that is currently being used. I just don't think there is a mechanism there for the LLM to self-direct or initiate anything outside of a session window, or outside of its token limit (which yes, is getting to be pretty large). If a new model is made that is constantly training while it's being used, and training on all its session tokens, then yes, that could potentially constitute a mechanism for self-direction. Personally I would think it would need more that just that, but that's where I would start drawing the line between static and active.

[DotArve]

My big issue with AI is that people are overly optimistic about it. When generative AI first showed up, and it still had some sort of novelty factor, it was fun. I used it to write me a Snake clone from scratch, just to test what it could do.

People seem to be thinking we're headed for some sort of AGI, limited or not - ChatGPT, Gemini and all the others are basically just less wrong versions of Eliza, and now to the point where they may occasionally be useful to explore new topics in which yoju have a willingness to do your own research, but they're nowhere near being "creative" - they're regurgitating a mashup of your input and the data they've been trained on.

The current iterations of AI are 50 years away from becoming AGI, and that's pretty much what we thought about AI in the 60s and 70s

[nataliealliepage]

Machine learning is fascinating stuff. Not all of it requires massive server farms, nor does it require mass scraping of the internet, nor does it completely eliminate opportunities for direct human input. I'm also fascinated by algorithmic art as a broader concept, things like Mandelbrot visual art or modular synth music.

[georgioszotos]

I don't need any of that AI. I got off the grid about a year ago. In fact, i made my own 150 wh. I don't need electricity to do my own cooking, laundry etc. I filled a cellar with old instruments, junk, pc's, and books to keep me happy building my own analogue AI, practising HI. I only left a door open for the outside AI. When this door closes, i won't jump from any cliff to find any friends. I will find them without a map. Even if i have to walk until the end. Then... bye-bye.

[youpattube]

I also have the stainless steel round magnifying fluorescent lamp that I see next to you. But your lense looks dusty. If you rotate the lense orientation by 90 d (the lense is perpendicular to the desk surface) when you are not using it, the dust just drifts on by and your lense stays clean much longer.

[Janokins]

I'm just waiting for the hype to drop out of this whole AI thing, like NFTs, I do wonder what the tech bros will think of next though
It's just too resource intensive, it often produces inaccurate results, there's a whole bunch of copyright issues surrounding it, and worst of all it was (in part) used as a justification for getting rid of me at the company. I used to use it to generate thumbnails, but now it just seems cheap and tacky. Now they (this agency) want me to use it to write my resume. I'm not being funny, but I'd rather have my own words on the page... or at least words written by a human. Surely resumes are read by machines these days to check if it was written by AI anyway.

[drivers]

nice segue "I can live with that... in the lab... it's where I work"

[RyanMacWee]

I hate the convenience of modern digital tech. Everything is immediately done for me, what is left for me to do? I’ve been getting into older tech and enjoy slowing down and savouring the moment instead of mindlessly consuming the lies of social media, getting shoved into echo chambers by “the algorithm”, seeing constant ads for betting services and cheap useless products.

And AI is just more “convenience” doing things for us, making us stupid cause we don’t know how to do anything anymore. Now we have more time to work and consume I guess.

[RoaldRoberts]

Born in '81, the year the first "modern" PC as we know it today was released by IBM. Automation Engineer in education (musician by heart) and know roughly 10 programming languages. I am probably of the last generation that has had to navigate a foreign country by physical hard-copies of maps, has had to use a telephone book, and am fine with the whole net going off and stuff and I really don't worry about AI getting out of hand. No doubt, we must be prepared to uphold ethics, moral standings and our society. But I sort of have this feeling, even if us silly humans create this thing that learns by itself, no matter what, it won't out-compete us, though it could compromise us. We don't know how to create an artificial brain at this point in time that is truly sentient, is what I'm saying. *And hoping* we won't for a long time, until we have a grasp on the subject because it apparently frightens people. It's good that it frightens us. This is why science fiction is good, it helps us go through scenarios in which humanity is in jeopardy. I think we'll be fine.

*BTW* I wouldn't say unstoppable, *inevitable* though.

[CedarCoveTigerPark]

A great book I read about this possibility is “When H.A.R.L.I.E Was One”. A look at what would happen when AI gains sentience, how it conceals the fact, argues for its existence, then offers humanity a token of peace by designing G.O.D. Highly recommend.

[alextirrellRI]

The only AI I've been enjoying is the one that lets me take fully mixed recordings and separate them into isolated tracks. I can finally remix a bunch of old live recordings of my band.

[Sir_Uncle_Ned]

I watched “Colossus: The Forbin Project” after a couple of bots developed their own language based on mathematics to communicate, Just like Colossus and Guardian.

As long as the first report from the first sentient AI isn’t those four words; we will be fine

[Skeptic]

Perhaps it's not the AI we should be concerned about. It's the decision-makers who determine who gets hired or let go.

[rothn]

This is kind of a weird take. I was going to cite some alignment research and speak to its goals and problem framing, but there are usually unexpected emergent properties of these systems at scale. Even in the latest game theoretic framing, one of the Nash equilibria is that the winning agent just _appears_ to be the best at satisfying humans. All of that said, I see this as far from the most likely outcome and I think it's worth trying.