Artificial Intelligence (AI) thread

kel

Ostrich
I work, somewhat, in AI. Most of what's marketed as "AI" would have just five years ago been called "doing a little math". AI is advancing, of course, but still I think the current AI trend will over time look a lot like 2010's "big data" trend. When was the last time you heard about "big data"?
 

Repo

Hummingbird
kel said:
I work, somewhat, in AI. Most of what's marketed as "AI" would have just five years ago been called "doing a little math". AI is advancing, of course, but still I think the current AI trend will over time look a lot like 2010's "big data" trend. When was the last time you heard about "big data"?

Not sure what you mean, big data is still super relevant, in fact what they call “AI” is usually used specifically on big data to analyze trends.
 

Blade Runner

Pelican
Let's just say I know a lot about the AI topic in medicine, particularly involving the medical specialties and sub-specialties. Let me cut to the chase of what's going on here, and yes, you'll have to trust me: patients will not get better outcomes (the only thing that matters). The marketers will do quite well with it, though, if they can convince the powers that be that it is good (for business, whatever that means). I think you can read between the lines here.
 

Repo

Hummingbird
^^. Can you expand more? Because some of us have lots of experience working with modelers and interacting with various forms of machine learning models, and my real world experience is the complete opposite of what you are saying. Granted, I’m in a different field.
 

redbeard

Hummingbird
Gold Member
What's the difference between "artificial intelligence" and "a fancy algorithm?" It seems to me that these are just man-made systems they tweak to become better at what it's doing. I'm not sure if it's really a step change in how we compute vs. a marketing ploy.
 

Aizen

Kingfisher
Artificial Intelligence and everything that surrounds it - NLP, ML, Data Science, etc - are some of the most secure jobs of the future economy. If you've got the type of mind that can handle this shit, I fully recommend looking into these career paths.

Once you get good at it, you can make a killer living, possibly even remotely. And there's plenty of jobs out there in these fields, so a man with integrity can avoid working for ((Big Tech)) or the billy goat gruff.
 

joost

Kingfisher
AI is the new 3D printer. Promising technology in theory but in practice is just another computation. Pure marketing words.
 

Stats

Woodpecker
I used to do research at a university computer science lab on ai about 15 years ago. Programing neural networks and teaching them by example various concepts and then comparing it to literature on how human infants learned those same concepts. A lot of what has already been accomplished in the field in the last years was thought to be imposible back then, never to be achieved in all future history.

Some of the biggest breakthroughs were when instead of just using one big neural network we hooked up several of them together much like how the brain is compartmentalized. And also when instead of just having a software neural nets we made super expensive neural chips which eche the neural net into the hardware of a physical silicon chips. last I checked just a handful of these chips where needed to be combined to reach or infinitely surpass the computational power of a human brain.

My understanding is that currently Ai outperforms human experts in pretty much every field if provided enough raw data to learn that specific field. But we are still a long way off from general super intelligence, where a ai can learn more then one specific specialty.

I think quantitative easing policy of central banks was largely to gain full control of all companies that are a the forefront of developing ai and to insure that they succeed. We can see how the fang stocks are pushing ai cognition. Tesla - computer vision. Google/facebook-computer speech/object/face recognition.

Soon we will have brain ai interface and then ai will be doing more and more of human thinking from that point forward.

I think where we are at a point currently where our present tech in ai is very difficult to distinguish from a real person when communicating with humans either through voice or txt. I think we are maybe 30/40 years away when ai controlled robots become indistinguishable from humans.

Conspiracy theory: its possible a super intelligent ai already exits either now or in the future. this ai is then influencing the past to insure its creation. using things such as the bitcoin mining network to exploit human gread to expand its computational capacities.
 

Blade Runner

Pelican
Repo said:
^^. Can you expand more? Because some of us have lots of experience working with modelers and interacting with various forms of machine learning models, and my real world experience is the complete opposite of what you are saying. Granted, I’m in a different field.

There have been posts on this before, but it is possible they are buried in other threads. Computation and cognition are totally different. I have spoken about flexibility (advantage human brain) and processing (advantage computers) as well.

Futurism is a tangent of scientism, which has roots of thinking which are similar to, or linking, evolutionary belief systems. For example, both use the faith statement "if given enough time" in order to try to convince you of XYZ, while there is only micro evidence that certain things can happen in relatively close or small states that we notice, going from one simple thing to another, even if effectively. The large, foundational or really complex realities that make us human and thus different than all the other creatures or creations don't necessarily follow this path, or are linked to it, at all --- at least in the way we think scientifically.

Perhaps it would be better if you asked specifically which subject you'd like me to treat.
 

Blade Runner

Pelican
redbeard said:
What's the difference between "artificial intelligence" and "a fancy algorithm?" It seems to me that these are just man-made systems they tweak to become better at what it's doing. I'm not sure if it's really a step change in how we compute vs. a marketing ploy.

redbeard does a great job of understanding the crude reality of AI, at its base level. You just published AI for dummies! Haha.

Philosophically, this is what an AI must be, since it is made by men.

I think maybe you have read my posts previously, although you didn't know it was me, because i found out long ago in really complex systems at least, it's all marketing. As a quick teaser, notice that in the videos they refer to "studies" and what or who did better at task X. Like statistics, you know literally nothing about what the format of the task was, how simple or complex it was, even if humans were compared to it --- or if that gauge or measurement is even meaningful in real life at all.

We have become accustomed to seeing technology work in our lives, and for those of us that don't have belief systems (that we're aware of) in the postmodern world, we as humans want to believe this is possible. Of course, the mass man of the 21st century doesn't have wisdom enough to understand that this is within him, nor that if he studied human history or the truths of God, he would already be aware or potentially discerning it when this passion arises.
 

Repo

Hummingbird
Blade Runner said:
Repo said:
^^. Can you expand more? Because some of us have lots of experience working with modelers and interacting with various forms of machine learning models, and my real world experience is the complete opposite of what you are saying. Granted, I’m in a different field.

There have been posts on this before, but it is possible they are buried in other threads. Computation and cognition are totally different. I have spoken about flexibility (advantage human brain) and processing (advantage computers) as well.

Futurism is a tangent of scientism, which has roots of thinking which are similar to, or linking, evolutionary belief systems. For example, both use the faith statement "if given enough time" in order to try to convince you of XYZ, while there is only micro evidence that certain things can happen in relatively close or small states that we notice, going from one simple thing to another, even if effectively. The large, foundational or really complex realities that make us human and thus different than all the other creatures or creations don't necessarily follow this path, or are linked to it, at all --- at least in the way we think scientifically.

Perhaps it would be better if you asked specifically which subject you'd like me to treat.

I realize that cognition is still a pipe dream, but I guess my point was even the advanced computations of machine learning is still a huge leap in many fields, and I’m unaware of any reasons that wouldn’t be the case with medicine. I have a hard time envisioning a future where your medical signs are taken and compared against huge sets of data and models, and not getting a better result than your local don’t-really-care doctor.
 

Blade Runner

Pelican
Ok, I perceive your bias, and I have no judgment on that per se. However, let's talk for a second about what I'm not addressing: concrete and indisputable algorithms of medicine. These can be helpful, adjuncts, or classification systems that are on the side or part of the EMR that can help. That's actually the first part of where AI should be applied to medicine. I don't want us to get bogged down here, but one really has to be clear and critical when he says, "better result" and "doesn't care" because they are qualitatively and quantitatively assessed; much subjectivity lies in those statements. And it might be that you believe "better outcomes" are out there, but they actually aren't.
 

Hypno

Crow
redbeard said:
What's the difference between "artificial intelligence" and "a fancy algorithm?" It seems to me that these are just man-made systems they tweak to become better at what it's doing. I'm not sure if it's really a step change in how we compute vs. a marketing ploy.

Think about sports betting. If you crunch enough data, it will tell you that homefield advantage is worth about 3.5 points. That's just regression analsis.

Low level AI will take the reams of data and massive computing power now available. It will tell you that variables you never even thought of matter to how you bet.

In the future, it will tell you these things before you ask the question.

The first couple of chapters of the book Homo Deus discuss the difference between humans and computers, and I recommend it. The differences are eroding and the trend is towards a type of merger.

A few decades ago people said a computer could never beat a human chess master. Then they said it could never compose a musical composition or art. Those differentiators are disappearing. At the same time, the study of human cognition has divulged numerous cognitive biases and blind spots and the frequent use of faulty heuristics.

So while some of the promises of AI may be exaggerated, out view of our own brains is also exaggerated. So the skeptics may be correct that real AI may be a mirage, AI that is better than the human brain is already here is you think about tasks in very discrete terms, and everyday it improves and becomes more broad.
 

Blade Runner

Pelican
I think we are now getting into threshold intelligence theory, and that is very specific to the task at hand. I am skeptical of AI for many reasons, but notice I only talk about it as applied to particular, high level things, not all things.

Things like chess, backgammon, etc. do have ends in brute force, by the way, so they are not the best examples, as they still fit the processing paradigm (computer advantage) more than the flexibility one.
 

Hypno

Crow
Blade Runner said:
I think we are now getting into threshold intelligence theory, and that is very specific to the task at hand. I am skeptical of AI for many reasons, but notice I only talk about it as applied to particular, high level things, not all things.

Things like chess, backgammon, etc. do have ends in brute force, by the way, so they are not the best examples, as they still fit the processing paradigm (computer advantage) more than the flexibility one.

You are correct, but at one time it was thought that they weren't susceptible to brute force.

Two corrolaries come from this. One, many more things are capable of brute force than we first thought. Things like art and music. Second, brute can result in things that look like imagination and exceed human imagination. Examples come from games like chess and go, where the computer has backtested unorthodox strategies and employes. The result is that the line between human and artificial intelligence is already very blurred and it will only become less distinct. The real opportunity is enhanced human intelligence.
 

Blade Runner

Pelican
Yes, you present interesting ideas which I am very open to. This is a very thoughtful thread, and one I look forward to revisiting as certain changes, innovations or claims come forth.
 
Top