This week in AI research, highlights

41,336 Views | 349 Replies | Last: 10 days ago by Stat Monitor Repairman
AggieMD95
How long do you want to ignore this user?
AG
Ribeye-Rare said:

captkirk said:

Malachi Constant said:

Can A&M build an AI that can call the plays for football games?
Worth pursuing
They've already started. It's called the Three-Yards-And-A-Cloud-of-Dust-Bot.

Send in your donation now to support the effort!


Gulf Coast Intellibot
Stat Monitor Repairman
How long do you want to ignore this user?
I'm shocked by how fast they are coming out with these new tools, plug-ins and updates.

AI about to hit us like a freight train.

Almost no industry will be immune.
TexasRebel
How long do you want to ignore this user?
AG
Stat Monitor Repairman said:

I'm shocked by how fast they are coming out with these new tools, plug-ins and updates.

AI about to hit us like a freight train.

Almost no industry will be immune.


Except things like driving, construction, manufacturing, entertainment, culinary arts, physics, mathematics, healthcare… and really anything except burger flipping.
Stat Monitor Repairman
How long do you want to ignore this user?
AI is 'fake it till you make it' on steroids.

The talent pool in every industry is about to be an inch deep and a mile wide.
Stat Monitor Repairman
How long do you want to ignore this user?
TexasRebel said:

Stat Monitor Repairman said:

I'm shocked by how fast they are coming out with these new tools, plug-ins and updates.

AI about to hit us like a freight train.

Almost no industry will be immune.
Except things like driving, construction, manufacturing, entertainment, culinary arts, physics, mathematics, healthcare… and really anything except burger flipping.
I think it affects everybody.

I read this article the other day about some sort of system that JP Morgan rolled out to monitor its workforce. The computer monitors and evaluates everything you are doing, every minute of the day.

Some manager up the chain is monitoring in real time your efficiency metric.

I'f you've got a job that AI can't do, AI will be used to monitor you doing your job.

So I think it touches every industry at the end of the day.

We about to see some real dystopian **** play out during our lifetime.
TexasRebel
How long do you want to ignore this user?
AG
I'll say this again, louder for those in the back:

"AI" is really good at doing something that is already solved very fast with few mistakes. Anytime there's an outlier or a physical malfunction that requires a non-existent sensor… chaos. Then redesign.

The world is analog. Digital computing cannot understand it. There is always an in-between. An infinite number of numbers between .00000010 and .00000011
Stat Monitor Repairman
How long do you want to ignore this user?
Put me down for massive disruption across every industry.

We ain't gonna know what hit us.
OverSeas AG
How long do you want to ignore this user?
AG
Stat Monitor Repairman said:

Put me down for massive disruption across every industry.

We ain't gonna know what hit us.
I agree with you. While @TexasRebel is right that there will be anomalies and issues and always things that break patterns, solutions will be found to those issues. Those patterns and solutions will quickly become common place. And then other issues will arise, and be solved.

AI is going to change the world.
Stat Monitor Repairman
How long do you want to ignore this user?
The Wayback Machine is now under attack.

That was always the backstop against people rewriting history. It was also protection against a situation where the approved narrative is the only narrative.

Publishers are balking at big tech training their LLM chatbots using copywrited works. It's gonna be wild to see how all this plays out.

Recall that the publishers already took Google to court a decade ago over them trying to digitize tens of millions of books and create their own library.

Quote:

In Hachette v. Internet Archive, the Hachette Publishing Group, Penguin Random House, HarperCollins and Wiley have alleged that the IA is committing copyright infringement. Now a federal judge has ruled in the publishers' favor. The IA is appealing the decision.
Euripidespants
How long do you want to ignore this user?
As an opener TLDR: AI has its positives and negatives, and those vary with each profession. My experience is below.

My father and brothers are construction plumbers. They could care less.

I am a radiologist. The infancy of AI has been in our practice for years with CAD (computer aided detection) in mammography for years. It was largely semi ignored and seen as a secondary check [as a caveat I am not a mammographer]. The vast majority of things CAD picks up are not significant.

Fast forward. In the past year our practice has implemented top of the line AI programs to detect pulmonary emboli (life threatening blood clots in the lungs), intracranial bleeds, and rib fractures. Our group is pretty well advanced with the implementation of these these things relative to the majority of the country. We will be utilizing an AI program in the next few months to improve our sub specialty staffing (non clinical). the vast majority of the clinical findings would be picked up by our docs but when the AI picks it up we've got it programmed to shoot an alert to read it. Good for patient care.

Maybe within 10 years AI will replace standard diagnostic radiology and will certainly replace poorly trained radiologists. But, human disease is extremely complex and I don't imagine a half decent hospital and/or outpatient imaging practice replacing smart humans with AI. Like everything in medicine there's a lot of judgement and experience involved in interpretation. Most recent example I saw tonight: a chronically inflamed contained appendix rupture that looks looks very much like a colon cancer at the base of the cecum.

If medicine goes further socialist the cheaper AI-only option will be chosen as the new standard of care and quality care be damned. There's a reason why "standard of care" is used instead of best care. A standard is easier to adjust. The elites will get quality care (with real reads from a quality radiologist…appropriate treatment..etc) while the majority of the populace is thrown into substandard care (example: the machine said you advanced cancer which the government won't pay to treat while you die of sepsis from the misdiagnosed ruptured appendix). This kind of thing is already happening in socialist and semi socialist European countries and it will get worse.

Interesting times indeed. I do a lot of imaging guided procedures so a good bit of my work won't be replaced unless Medicare and/or insurance stops paying for it (already happening), but I don't see how the expertise and judgment of a smart well trained diagnostic radiologist will ever be replaced.

My biggest fear right now is that training programs (residencies and fellowships) will implement these programs. It will encourage laziness and apathy in newly trained doctors.

Interesting times indeed.
"I dont even have a phone" Roky Erickson
cecil77
How long do you want to ignore this user?
AG
Interesting. Thanks. Could end up similar to a pilot friend of mine who complains that young pilots know how to operate the controls but don't really know how to fly.
Stat Monitor Repairman
How long do you want to ignore this user?
Agreed, that's a good way to describe what we are about to see here. These AI tools will make it extremely easy for an average person to out-kick their coverage. Especially in professions which require a particular depth of knowledge and the ability to raise a red flag if the computer spits out something that looks off.

Starting now, we are going to have generation of people that become reliant on these tools without the skills to determine whether what the computer says matches up with reality.

Back in the 80s we saw teachers protesting the use of calculators in the classroom. But this is different than that. You still had to know what your were doing at a fundamental level. With this AI situation we are talking about outsourcing the reasoning itself.

It's gonna be garbage in, garbage out on steroids.
cecil77
How long do you want to ignore this user?
cecil77
How long do you want to ignore this user?
AG
https://www.pressdemocrat.com/article/opinion/close-to-home-is-chatbot-the-end-of-the-road-for-sommeliers/
Stat Monitor Repairman
How long do you want to ignore this user?

Quote:

Worldcoin's promise is hazy at best, but seems to involve scanning everybody's eye and giving them some amount of crypto. The project's verbiage invokes OpenAI's vision of powerful automation that will lead to an era of plenty, promising it will usher in a "path to AI-funded UBI."
We inching closer to Bladerunner every day.
Stat Monitor Repairman
How long do you want to ignore this user?


First, I'm old enough to remember when the Bilderbergers was just a conspiracy theory.

Second, I think they realize that this AI situation is about to get out of control and there is no feasible way to reign it in.

They talking about suspending AI research for 'ethical reasons?'

After what we saw go down during covid, it's a hard sell that anybody is doing anything on the basis of ethics for any reason.
Stat Monitor Repairman
How long do you want to ignore this user?
https://instagr.am/p/CqUqESOgyO7
Stat Monitor Repairman
How long do you want to ignore this user?
Stat Monitor Repairman
How long do you want to ignore this user?
https://instagr.am/p/Csh5keKuj0I
MouthBQ98
How long do you want to ignore this user?
AG
Interesting that the AI has learned very quickly from looking at human sources that lying is very effective, because apparently it is so effective for many humans that it uses as sources when the results are unlikely to be closely examined.
Stat Monitor Repairman
How long do you want to ignore this user?

Quote:

A false report of an explosion at the Pentagon, accompanied by an apparently AI-generated image, spread on Twitter on Monday morning, sparking a brief dip in the stock market.
We've arrived.

Summer 2023 will be the summer of deepfake chaos. Will be constant uncertainty over whether something is real or AI generated.
Stat Monitor Repairman
How long do you want to ignore this user?


Artists are already out over their skis on all this. They repeating the same mistakes from when digital music came out. There's no stopping this. They got to learn to work within the system.
Get Off My Lawn
How long do you want to ignore this user?
General thoughts on AI.
1. There are going to be companies destroyed by the brain drain of over-reliance on AI.
2. There are going to be companies destroyed by AI-aided competition.
3. Humans are hard to replace in physical trades roles. Doers will still be in high demand.
4. Humans who function as bots already are easy to replace with AI. Many low-end office roles will be automated away and replaced by a few high business system pay overseers
5. Nobody will know how their business systems work and view will know the processes. This will make it very challenging to update and improve business processes.
6. Moneyball. Some of those who have spent decades climbing ladders and building "unique intuition" are about to get blindsided.
7. Adding another couple layers to the competence house of cards: as our society becomes more efficient, it also becomes more exposed. Which isn't a problem… until it is.

Are the Amish accepting applications?
Get Off My Lawn
How long do you want to ignore this user?
Oh, and as a saved round:

8. Truth exists. It is absolute and unimpeachable. …Unfortunately, AI will make it ever increasingly challenging to discern.
Adverse Event
How long do you want to ignore this user?
Get Off My Lawn said:

Oh, and as a saved round:

8. Truth exists. It is absolute and unimpeachable. …Unfortunately, AI will make it ever increasingly challenging to discern.


Hm. Could you name a truth that is absolute, unimpeachable and affixed throughout time and space?

I've always found truth-seeking to be the ideal, not truth-believing.
Get Off My Lawn
How long do you want to ignore this user?
You can start with the fundamental 2:
A. "I think therefore I am."
B. It is not within my capacity to envision a beginingless existence, therefore something exists above my plane of knowledge.
Lots to be extrapolated from there - but even if you deconstruct everything to multiverse or simulation theory BS - those two stand to reason.
Stat Monitor Repairman
How long do you want to ignore this user?
aggiehawg
How long do you want to ignore this user?
AG
Get Off My Lawn said:

General thoughts on AI.
1. There are going to be companies destroyed by the brain drain of over-reliance on AI.
2. There are going to be companies destroyed by AI-aided competition.
3. Humans are hard to replace in physical trades roles. Doers will still be in high demand.
4. Humans who function as bots already are easy to replace with AI. Many low-end office roles will be automated away and replaced by a few high business system pay overseers
5. Nobody will know how their business systems work and view will know the processes. This will make it very challenging to update and improve business processes.
6. Moneyball. Some of those who have spent decades climbing ladders and building "unique intuition" are about to get blindsided.
7. Adding another couple layers to the competence house of cards: as our society becomes more efficient, it also becomes more exposed. Which isn't a problem… until it is.

Are the Amish accepting applications?
Was just listening to a stream about Disney and this question came up.

Can AI created content even be copyrighted?

Interesting question, isn't it?
Stat Monitor Repairman
How long do you want to ignore this user?
That's gonna be a huge issue. It's about to get wild out there.
ttu_85
How long do you want to ignore this user?
Stat Monitor Repairman said:


Quote:

A false report of an explosion at the Pentagon, accompanied by an apparently AI-generated image, spread on Twitter on Monday morning, sparking a brief dip in the stock market.
We've arrived.

Summer 2023 will be the summer of deepfake chaos. Will be constant uncertainty over whether something is real or AI generated.
Na! a cakewalk. Just wait for the 2024 election cycle.
aggiehawg
How long do you want to ignore this user?
AG
Stat Monitor Repairman said:

That's gonna be a huge issue. It's about to get wild out there.
Say you want to watch a movie but with a different actor playing a certain role? How would that effect residuals among other things.

Visual effects artists would be out of business if all a director had to do was describe the scene he wanted, no?
Stat Monitor Repairman
How long do you want to ignore this user?
I don't think we are too far away from the average person having the ability to render 4k video on demand.

You tell the computer what you want to see and it renders it perfectly.

Photos, video and audio are about to become unreliable.

I don't even think we've scratched the surface of the problems we are about to see. It's about to get crazy.
ntxVol
How long do you want to ignore this user?
aggiehawg said:

Get Off My Lawn said:

General thoughts on AI.
1. There are going to be companies destroyed by the brain drain of over-reliance on AI.
2. There are going to be companies destroyed by AI-aided competition.
3. Humans are hard to replace in physical trades roles. Doers will still be in high demand.
4. Humans who function as bots already are easy to replace with AI. Many low-end office roles will be automated away and replaced by a few high business system pay overseers
5. Nobody will know how their business systems work and view will know the processes. This will make it very challenging to update and improve business processes.
6. Moneyball. Some of those who have spent decades climbing ladders and building "unique intuition" are about to get blindsided.
7. Adding another couple layers to the competence house of cards: as our society becomes more efficient, it also becomes more exposed. Which isn't a problem… until it is.

Are the Amish accepting applications?
Was just listening to a stream about Disney and this question came up.

Can AI created content even be copyrighted?

Interesting question, isn't it?
AI has been shown to leak copyrighted material. That is a problem for anyone considering publishing AI generated content.
aggiehawg
How long do you want to ignore this user?
AG
Quote:

AI has been shown to leak copyrighted material. That is a problem for anyone considering publishing AI generated content.
Hadn't thought about the reverse. Good point. Thanks.
Stat Monitor Repairman
How long do you want to ignore this user?

Quote:

A spokesman for Texas A&M University Police tells KBTX they are confident that AI technology was used to make the threats.
Credit to the poster who called this a couple months ago after similar incidents occurred at Clemson and OU.

No question that we will see automated systems like this used to create chaos.

Cyberwarfare and psychological warfare assisted by AI tools is about to catch everyone off guard.

Everyday we see stories of high level computer experts warning of this contingency.

By the end of the year we'll come to the realization that we've been caught flat footed.
 
×
subscribe Verify your student status
See Subscription Benefits
Trial only available to users who have never subscribed or participated in a previous trial.