Effects of Generative AI - Big And Small?
 

This thread is to speculate about how generative AI will be used and change things, in ways big or small.

Subscribed and interested.

I can't wait to see how the big thinkers of the board futurescape our brave new world.

Here’s a “small” effect - I think more people will go to Youtube to learn how to do things or figure out what to buy.

A Google search for, say, what ski goggle to buy is rapidly becoming useless, there are dozens if not hundreds of articles and reviews and buying guides on the web that have pretty clearly been cranked out by an AI or with assistance from one. There aren’t AI Youtubers, for the most part, yet. You can look for sources that you recognize as actual experts, but with the death of magazines, most of those expert sources are going to be on Youtube. In this sense, I think it’ll be good for Google (owns Youtube).

I posted a thread in PARF but it got lost in the real outcome.

It'll replace google because Americans are lazy and cannot decipher fact from fiction.

OpenAI is very good with grammar and readability/Flow. The PI I work under uses it to polish his grant proposals, LOR, ect. I have used it for the same and so far have been awarded two grants. ;)

The problem is who owns the writing. If I tell it to polish paragraphs. The material I gave it is mine, but changing a couple sentence structures and punctuation shouldn't be considered plagiarism...

With direct Q&A content. It gives very good answers, and a logical process of how it came to that conclusion. Often the answers are wrong though...

I see AI taking over a lot of jobs that just require reading email or answering a phone, looking up info, and spitting that info back out. (Customer service, order entry, etc). Probably already there...

Is still early dayz. I think what we have now from ai is raw id with no reason. They can’t count, they can’t solve problems. They’re ‘stochastic parrots’.

The chatgpt front ends are now being attracted to problem solvers and we’ll end up with things that can find creative solutions to actual problems. Doing taxes, finding tax cheats, planning supply chains, selling stuff.

Worst case: current chat front ends will be trained to advertise and convince people to buy and do stupid stuff. A small elite will benefit.

At this point it seems more like Google with good grammar. I've asked quite specific questions and the responses reminded me if you asked a politician to speak for 5 minutes on a topic they know nothing about. The words all fit, but your question is frequently not answered.

I think it has promise to take over some mundane tasks and actually benefit people. For example an AI assistant that can help me with my work calendar, answer repetitive questions that people ask, etc...

My concern is the generators. Will they essentially fill the internet with fluff making it harder to actually get in depth information as its lost in the sheer volume of AI content.

Like you said, there are a slew of sites that "review" or "compare" products, but then when you get there, they have a list of the most popular stuff with a short list of the features as advertised by the manufacturer. Those sites tend to say "product X is the best", but there's no actual testing or reviews other than having looked at the features. Youtube is also full of videos of the same information using a computer generated voice. I have been assuming it was folks in other countries, but it's absolutely possible that those folks are using some sort of AI to create the videos.

AI generated Sci-Fi is flooding publishers to the point that they stopped accepting submissions.
https://www.theverge.com/2023/2/25/23613752/ai-generated-short-stories-literary-magazines-clarkesworld-science-fiction

AI can be trained, not only to make art, but to mimic a certain artist's "style". The US has said "you can't copyright AI generated content" but does that keep someone from using to make money it even if it's not copyrighted.
https://www.theverge.com/23444685/generative-ai-copyright-infringement-legal-fair-use-training-data

I love these emimen AI generated voice videos . There are dozens of them .
https://youtu.be/wUHf7wUYHvQ

I'm unapologetically a fan of pulp fiction westerns. I think AI could probably write as well as many of the authors that I've come across. But, I wonder if in the future, you could plug in a few parameters and have the AI continue a series of stories from an author, 'in the style of' that would be almost indistinguishable from the original author's style. Imagine AI continuing stories of Travis McGee or Spenser (for hire) Not written by McDonald or Parker but good stories nonetheless with the original cast of characters.

Here’s an example.

Like many professionals, I often get unsolicited emails from headhunters. Typically, these are from human beings who’ve been searching LinkedIn and sending out emails to likely candidates.

Last week, I get this from someone named ‘Joey’.

The signature to the email was also just Joey. No company name, no phone number or website.

I checked out the company that was ‘recruiting’ me and saw they don’t have a presence in the NY market. So looking at the email it was sent from, I found this joey@findemtalent.com.

Searching for that domain, I came upon a company this company https://www.findem.ai/solutions/platform that evidently uses AI to solicit prospective employees. Likely capable of doing the work of hundreds of humans in a fraction of the time.


Sent from my iPhone using Tapatalk

<iframe width="1520" height="585" src="https://www.youtube.com/embed/jPhJbKBuNnA" title="I tried using AI. It scared me." frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>

This is a hint of a huge change coming....

People have already run into that to some degree. Living artists are not terribly happy about having their style copied, but in original work. And the way that it's done today, is that electronic copies of the author's original work are fed into the AI, and once it's digested enough original work, it can emulate the style in new work. So, while it may be "original" from a certain point of view, it was only able to emulate that work after being fed a bunch of other work and then being asked to essentially copy that style. So I can see how there would be an element of "theft" or "plagarism" that doesn't really fit with the old/current definitions.

What about AI enhanced deep fake videos? Could that be a thing?

I think they’re already here.

Absolutely. We're probably just a couple of years away from deep fakes being undetectable...at least without some effort.

It has been for years.

<iframe width="860" height="615" src="https://www.youtube.com/embed/GbIk5GXhgUk" title="YouTube video player" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>

Even 15-20 years ago I though Hollywood movie scripts could be written by a computer.

It’s just a bunch of the same tropes and cliches strung together for 90 minutes.

AI will easily be able to write equal or better scripts than 90% of movies made today. It probably already can.

Also for movies, it seems like they way they are produced is going to change dramatically. All or mostly all done on a computer.

But overall I think it is going to be a huge impact on society. I’m not sure of all the ways. But some obvious ones are anything involving written work. For example, legal briefs. I’m guessing what would have been 30 hours of billable time by 2 lawyers (a senior and a junior one) to do standard-ish law and motion can probably be done by 1 in a few hours. AI, especially in conjunction with a specially developed program, can probably get it 90% of the way there with a push of a button.

So a huge thing is how it’s going to impact employment and the workforce. Seems like it’s going to eliminate a lot of jobs/man hours, and fairly soon.

Answering the phone will be the next big thing for customer service calls. Instead of calling mega corp and the call routed to India and a person with an accent so heavy that can't be understood, AI will be answering the call, and it will have to figure out human accents and try to solve problems.

It will still be a few years out before AI has learned enough accents and speech patterns to understand the questions. I was in Plymouth Mass. and trying to communicate with a parking lot attendant and I only understood only every other word at best. I finally threw up my hands and told him I was just going to turn around and leave and he nodded. His accent was just gibberish to me.
AI will have a hard time with it as well.

AI is mostly just copying what humans are doing, so AI is not likely to be much if any better, other than being able to turn out new versions of the same old thing at a quicker pace than a human.

Major implications.

Global and pervasive.

It all started here:

https://home.dartmouth.edu/about/artificial-intelligence-ai-coined-dartmouth

https://www.forbes.com/sites/gilpress/2016/08/28/artificial-intelligence-defined-as-a-new-research-discipline-this-week-in-tech-history/?sh=3efb8d336dd1

I had classes in Dartmouth Hall many years later and I never knew.

Dartmouth Hall has been recently renovated... alums donated $26 million toward this effort.

There is more than one connection between Dartmouth Hall, the origins of AI, ChatGPT and OpenAI.

At least one is a key businessperson.


Coming soon: Gort.

Laurie Anderson was on the Maron podcast discussing a company that (A) transcribed the Bible digitally, (B) transcribed all her work digitally, then ran a program thru GenAI that rewrote the Bible in her words and thoughts. Then, they (C) ran the Anderson Bible through the collected works of Lou Reed.
I haven't read any of this results so I have no opinion on the final project, but she seemed delighted.

Maybe more and more mashups and projects like that?

Mein Kampf as translated through Gandi's teachings?
Zane Grey influenced by Bukowski?
Curious George as written by Jerry Garcia?
A movie like Wizard of Oz as written by the Coen brothers?

A technical manual written by [the engineers who developed the widget in question] re-written in the style of Dave Berry?

“Our findings indicate that approximately 80% of the U.S. workforce could have at least 10% of their work tasks affected by the introduction of GPTs, while around 19% of workers may see at least 50% of their tasks impacted. The influence spans all wage levels, with higher-income jobs potentially facing greater exposure.”

https://arxiv.org/abs/2303.10130

This is just one of many studies rushing to assess turn potential impact of generative AI on jobs. I have little faith that these early efforts are anywhere near right or that their methodology is sensible at all. I’ll bet a lot of this is just academics rushing to build a reputation in a hot new area. But they could be wrong in *either* direction.

Best case, AI makes people more productive in their work. Worst case, AI replaces people. In most cases, I think a combination: people become more productive, so you need fewer people, even if AI isn’t doing the job all on its own.

A lot of higher education, higher pay jobs seem quite ripe for AI productivity boost and hence headcount reductions. Thinking back to my lawyer days, AI could have significantly sped up legal research, document review, drafting briefs. As a junior equity analyst, a lot of what I did is stuff that AI can do or will soon be able to do. For what I do today, AI can eventually be a big timesaver. The quant and algorithmic trading shops are quick to adopt any new tool, and have been using other forms of AI for some time.

To be clear, I think there may be not that many jobs that can be done *entirely* by AI, whether generative or other. But I think there will be more and more, and ultimately quite a lot, of jobs where AI can do part of the job with humans using the AI output, selecting from choices prepared by the AI, or supervising the AI. Fewer humans. And eventually, humans who never learned to do the work now entrusted to the AI, so that more and more of the knowledge and skills required to run the business or operate the equipment exists only in the AI.

Where do you get experienced CPAs from, if 50% or 80% of the junior and trainee CPAs no longer exist? Ditto radiologists, pilots, insurance agents, lawyers, programmers, librarians, etc.

I’m not liking where this is going.

This is exactly what we are doing right now. We are currently using AI (machine learning, more accurately) to optimize the control system for our supercritical (cryogenic) hydrogen system. This system is really one-of-a-kind in the world and the controls architecture needed to maintain temperatures and pressures during thermal transients is very difficult to manage (or troubleshoot when things go wrong). When "things go wrong" we experience a pressure excursion that blows rupture disks and vents hydrogen to the atmosphere. We embarked on an AI-ML project to study the system and it ended up spitting out some PID control parameters that were wildly different than what we were previously using. We loaded those parameters and tested the system under controlled conditions and it performed wonderfully.

We would likely never have gotten there without the ML effort.

This Robot Already Owns Everything (And it's just getting started) : Blackrock Aladdin

Blackrock has a secret weapon that has made it the most powerful company in the world: Aladdin. If you're ever wondered how Artificial Intelligence could impact our lives, here's the answer. Aladdin is the brainchild of Larry Fink, and it already controls more assets than the GDP than the US. It's growing by $1 trillion to $2 trillion new assets in its control each year.

https://youtu.be/AWBRldjVzuM

I saw an announcement from Microsoft that useers of Office 365 will soon have IA spell check and other functions soon. Right now only the Enterprise version has it, but it will soon be added to regular peons like myself.

I can see that as good and bad. Good to have better spell and grammar check, but how much of that text, (likely all) does it catalog and store for it to sift through looking for a subject or combing through for words and phrases that the programmers deem "bad" or racists or against the agenda of the programmers.

Anyone read about Alpaca software? AI software that trains itself. As explained by Dr. Know-it-all on YouTube this is a game changer.
You can Google this to get to the video. Start at 5:09.
Moody's UPGRADES Tesla Stock! �� And ChatGPT Learns to Train Other Large Language Models!! ��

The movie ‘Close Encounters…’ scene where the terrestrial computer and the alien computer begin to learn how to communicate with each other with ever accelerating speed to quickly take control of the process was actually, IMO, the most intriguing and scariest part of the whole film.

Yes, and it is happening now. I don't think we are able to imagine where this will go.

I agree. I think 10 years from now the impacts of AI on society will be immense. Whether in a positive or negative way remains to be seen. Kinda like the internet...it's a powerful too that enables the good and the evil at the same time. I suspect AI will be similar.

I have a bet going with my wife that lawyers will be replaced by AI/machines before burger flippers. It's a race between machine learning and robot dexterity, I think. But that's generative AI, the next step is general AI.

With general AI every job can be replaced by a mixture of AI and robotics. This isn't like the industrial revolution at all, where whole new fields of work opened up when previous ones closed; every new job created will be replaced, at some point, by AI, until such time that new fields are limited only to AI.

We're 30-50 years away from that sort of general AI, but it will come.

What will be left to us? The purely theoretical? Brand new concepts in math and physics? Maybe for some tasks the energy requirement will be too high for AI and it will let us handle them because it's cheaper to do it on caloric intake in lieu of electrical power.

Back to your point, I hadn't really considered the training aspect before. It's interesting but in the short to medium term I think a shift in education could train new experts who are aware of and understand the basics now handled by automation but who focus, from the start, at the higher level management functions.

The question, I think, is how quickly education can keep up with the advancement of AI. ChatGPT went from a 10th percentile to a 90th percentile on the Bar exam in a year. Dall-E was an interesting little low-resolution picture generator in 2021, now AI art is replacing real jobs and nearly able to generate believable video from text.

How does college prepare someone with a four year degree when AI advances at the rate it does?

Edit: I think the biggest question underneath all of this is is how do we guide nascent AI? There's no stopping it, that's for sure. No matter what regulations are put in place to protect this or that industry they'll be only temporary at best. I feel like we're fully aware of the onset of nuclear weapons fifty years ahead of time, and nobody wants to talk about the rules of engagement.

Originally Posted by Z-man View Post:

‘In theory, yes - it is pretty interesting stuff. But practically speaking, an AI computer is essentially a relational database with computational algorithms that can make connections via the database and generates output based on certain criteria. Yes, a database can grow (provided there is sufficient storage available) with more input, but the computational algorithms can be a bit difficult to grow autonomously to the point where increased intelligence (or rather improved algorithms) can be measured. Therein lies the rub.

Too many people confuse the issue by perceiving a computer mimicking human behavior as intelligence. The computer is only following a set of algorithms, which is not what intelligence is.

I would not worry about a Terminator knocking on your door just yet...’

-Z



The above quote is from 8 years ago.

IMO, we are getting real close to worrying about a Terminator not necessarily knocking on our doors but rather locking us out of our own lives.

Seriously. Millions of us have our fingerprints on file someplace, maybe as a requirement to own a handgun or our parents trying to protect us when we were kids from abduction, or even because of having been arrested.

If we expand that beyond fingerprints to include facial recognition, biometrics, our entire digital histories, medical histories, financial histories, family histories, our properties, our everything about us…you get the picture.

Let’s say all that stuff about all, or nearly all 6 billion or whatever of us, is stored in a digital box. Now let’s say some evil genius like the box labeled Bill Gates decides the box labeled ‘Crowbob’ no longer has any relevance.

Poof, gone.

Alive but forgotten, locked out. ‘Access Denied’ to everything.

Once you realize the real Bill Gates had already passed away years ago but the Bill Gates box ‘lives’ on is not a comforting thought.

Who wants to hear Steve Jobs' opinion on Covid, and current AI development?

https://twitter.com/BEASTMODE/status/1637613704312242176

The question is at what point do neural networks "think" better than humans?
E.g. they trained a NN to play Go. It got better and better until it crushed any human player. The style of play was not recognizable. The NN did not have to be programmed - it "programs" itself. This is the "issue". Now you take two NN and let them play each other. The teacher NN is better than a human and the student learns until it gets better than the teacher and then they change sides. I don't think we know how this will turn out. What happens if one of these NN gets access to the internet with nefarious intentions? What if NNs are tasked with playing the stock market?

Doesn’t BlackRock already use some kind of algorithm which some think is AI?

My concern,
Personalization that recognizes your buttons and alters your inputs to push those buttons.
Result? Increase societal polarization, perhaps even to the point of violence.
This may already be happening now with places like youtube.
However, the next step is not as much gatekeeping what you can see, but tailoring what you see.
Two people could que up the same show, but each might see something with AI altering the words or visuals in places.

On the good side:
There currently exist people that come up with good ideas but have a rough time filling in the details.
AI working for these people would bring about an amazing invention rate.

On the flipside, there are also people that are going at taking a concept and working out the details, yet lack conceptual spark.
An AI could spit out concepts, and the detail oriented people could pick something that interests them and make it work.

My concern is that what little justice we have now will go away completely. How do you punish code?

Who will oversee AI if AI is overseeing us?

I think if not already, AI will soon be out of the bottle and won’t be put back in.