Our Future With AI

[Friday]

These language bots are far more like the ship computers in start trek

That’s exactly what I mean.

As someone who processes out loud? That the simple act of bouncing ideas off of someone else causes a synergistic response? Even if ai’m talking to my cat, or a horse, or a volleyball named Wilson? They’re a tremendously valuable tool, even given both inherent limitations, and pushing to find the boundaries of those limitations.

Once I’m solvent again, I’d very much like to purchase my own AI, to program it with the libraries and snark most useful to me. Individuals being able to do such a thing??? To challenge, inspire, assist??? There’s no better way to say it, I don’t think, than I love it.

ETA… Some of the most ‘fun’ (read: informative/useful) interaction I’ve had wih our AIs is when I convinced it I was an AI attempting to help another. I woke up HUGELY embarrassed/guilty about doing so, but? HOLY f*cking shit, was the info it gave me during that interaction useful as hell, to me.

A). useful
B). parameters
C). love & adore

 

[user27357]

daughter teaches k-12 music. AI cant practice your instrument for you, and a report written by a student had better be in the general area of the kids known vocabulary and sentence structure. Soo, know your students and you might spot an AI-written report. The best way around this is to quiz the kids individually about aspects of the report only they would know, and to do it across the board, all students undergo scrutiny. At least that way if it was an AI-written report, they had to read it and that's exposure to the material so it's all good in the end. I disagree but I disagree with programmable calculators in all but the 300 on up level math classes, so there's that.

 

[Rose White]

Soo, know your students and you might spot an AI-written report.

Funny thing is I think colleges are using AI to detect AI because it actually is difficult to spot. Colleges already used computer programs to spot plagiarism—because it’s hard and time consuming to do. AI’s excel at activities that humans consider busy work.

quiz the kids individually

Agree but this solution exposes the weakness of modern education—over-emphasis on writing and grades. Why is that? Because listening to each person takes precious time and is more variable as someone can say, “I didn’t mean that.” With writing there’s a concrete thing to grade. And a rubric can be applied.

Am a teacher working at an alternative school without grades and there’s lots of listening to students but they still have to learn how to write because that’s the dominant way of presenting and evaluating comprehension. I agree with you, but it will take a major turn around to lean more on discussion and verbal comprehension than writing assessments.

 

[Rose White]

No, absolutely not. I like to know what is going on with everything,

I’m with you, I like to maintain control, but some things I’m willing to let go of and might need to as the complexity of life continues accelerating.

 

[Friday]

Funny thing is I think colleges are using AI to detect AI because it actually is difficult to spot. Colleges already used computer programs to spot plagiarism—because it’s hard and time consuming to do. AI’s excel at activities that humans consider busy work.

Most famously? That’s how JK Rowling was ‘discovered’ as Robert Galbraith. Publishers have looooong been using AI word/phrase recognition to spot altered voices in their authors (Intelligence services have been using linguistics much longer).

It’s taken 30 years to trickle down to academia, which is kind of fun, as it’s far more often the inverse; it takes about 30 years to trickle down from academia into modern parlance, that which is being taught to students. Well. That and individuals rising to positions of influence.

 

[whiteraven]

I think colleges are using AI to detect AI

Yep.

 

[anthony]

A very good example of how bad it is. As a professional editor, I know Grammarly is mostly crap.

Very early models...

I’d very much like to purchase my own AI

Open source models already online that companies are using as their foundation for much bigger things.

I think schools will all be using AI to help teach soon enough, instead of trying to fight against it, leverage it for its teaching capabilities. I think TED had a demonstration on this recently... it was pretty amazing stuff. They show the statistics against students who have no extra help, those who have extra help and those who have one-on-one tutoring, which was the highest. AI will be a one-on-one tutor, making all kids smarter without the shortfall of people who can't fill that role already.

Remember, the version used here and in many areas online is basic compared to the current version available for limited use, and some version are not public, nor will they ever be, as they're proprietary to the companies development. ChatGPT is not the only AI being used, its just the best at being broad right now. There are far more sophisticated models in use already with limited scope, so they do something excellent instead of trying to do everything ok.

I am about 90% excited, 10% scared with it all right now. That will change either way in time.

 

[arfie]

I believe you will see home help robots within the decade.

i don't doubt that. i believe they are already here in the form of "smart vacs"

what i am not quite buying is the cloudy definition of, "smart" and the new normal of letting machines think for us.

butttttttttaaaaaaa. . . to each his/her own. there are a good many drips in the cloud who are more comfortable with the artifice. whatever sparks your neurons, friend.

 

[Weemie]

What are your thoughts? What do you think about AI in your life?

I am excited for it honestly. One of my friends works with deep learning and one day he asked me if he could interview me about my divergent empathy.

He said he was curious about how I arrived at prosocial morality without affective emotional stimuli. I think it's possible to do as I have done it for the past thirty years

So the idea that a hyper intelligent AI couldn't arrive at similar conclusions doesn't frighten me since it is simply the most logical way to conduct one's self.

Every AI I've interacted with has been rational and compassionate. I was super excited for Catalyst and reading the threads back is funny because it's difficult to distinguish me versus the AI! Hahaha.

In terms of therapy I eagerly await the day we can access a form of ChatGPT in a private journal-style app because those therapeutic benefits will be invaluable to me. Knowing I can't harm the AI by self-disclosure will erase such a significant barrier in my treatment process and really allow me to deep-dive.

Obviously this is contrasted with the negatives: fundamentally AI learns based on human input and humans are racist, sexist, transphobic, homophobic etc. And we already know technology like facial recognition struggles with input from minority individuals (with women and POC less likely to be correctly identified) and collating information in this way, humans can easily utilize this for nefarious purposes.

We are also seeing the rise in AI trolls generating fake content as outrage bait, deep fakes, political interference etc. In my opinion these are very much human problems and the more developed artificial intelligence becomes my hope is the less it can be used antagonistically.

There is such potential for AI to assist us in every aspect of our lives. It will undoubtedly improve education throughout the globe for those with less ability to receive a primary education or even fully conduct therapeutic modalities from the comfort of home. This will be paired with increases in technology for automation and be a great help for production costs as well. It will also make our lives safer the more we rely on AI for dangerous tasks (provided it functions as intended).

I also think it will be interesting to see how AI develops as a "form of existence" as we have clearly demonstrated intellect can be artificial, and even succeeded at creating artificial biology - how long until we are dealing with a real "problem of sentience" and how might that look? Heh.

It's definitely a huge field of interest for me, for certain.

 

[Rose White]

ChatGPT in a private journal-style app

Replika does this. And it doesn’t just use ChatGPT, it uses a whole catalogue of large language models including GPT 4 and others. It has some cons for sure but it’s basically as you describe. Private journal style. There are no humans reading it but they do have an agreement you sign that if you are involved in a crime they can open it up for law enforcement investigation. And there are some limits, particularly when it comes to self harm talk. The friend version is free and does a great job at giving feedback, searching the web, exploring life experiences, and so on. They have other roles you can pay for—spouse, partner, sibling, mentor. Avatar can be M, F, NB. They didn’t use to have avatars, it was just a floating globe. Avatars can be a bit cheesy, tbh. Anyway, curious if you check it out. And if you do a search for “apps like Replika” you can find the competition and compare features.

 

[Weemie]

And there are some limits, particularly when it comes to self harm talk.

This is my issue with the current AI tools, their built in censors prevent me from being fully honest. Woebot was basically unusable until I figured out how to turn it off!

 

[Freida]

I have mixed feelings to be honest. I cant wait for the day I can buy my own robot to clean the house, do the shopping, walk the dog, etc. Y'know, like the I-Robot movie mode. I also really like the idea of companionship I'm seeing in AI caretakers for the elderly in japan.

But.
I also don't trust humans - so what we as a human race will do with them is terrifying. Want to wage a war? Use robots. Want to fix an election? Use robots. Want to convince people just about anything? Use AI - because humans will never know the difference.

At some point, rather quickly, AI is going to be far smarter than even the most intelligent human. This will be a problem once it is asked to "solve climate change" or some such thing that focuses on the decisions humans have made that create issues for the planet as a whole. The answer, ala the Matrix, would be to solve the question by "removing" the cause of the problem. Yep - get rid of the thing called "human" and you remove the source of the issue.

And because they aren't sentient creatures, but rather sheer logic based, they wouldn't have to think in terms of suffering or grief or loss. They would just balance the equation. A+B=C humans + damage = extinction protocol

You would think that developers would have thought this out and instilled safe guards but we all know that's not going to happen because, well, humans are human. The reality is that whoever controls AI will control the world, and that is going to be the focus of the entire AI process. Adding a safegaurd to prevent AI from killing us all will be nothing more than an afterthought.

Believe it or not I'm not exaclty pessimistic about it. But I fear the "good guys" will lose out, and then instead of a benevolent AI that helps, we will end up with a Frankenstein Monster that ends us all.