Science/Techno Babble Random, Random

All the other crazy stuff we talk about. Politics, Science, News, the Kitchen, other hobbies.
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#541

Post by Suliso »

ponchi101 wrote: Thu Dec 22, 2022 12:32 am I understand that Climate Change is one very serious threat to humanity. But not far behind (and I do not know if they are behind) stand Work Automation and Artificial Intelligence.
Here is one example:
AI can now write like a human. Some teachers are worried.

The Gist: A company called OPENAI has released a language model that is so complete that it can produce essays that are impossible to tell from human-produced ones. Meaning that, in a short while, every idiot in school could deliver an essay that the teacher could not tell it would be produced by AI.
Imagine the consequences for any and all writers: newspeople, scripts, novelists, anybody making a living by writing.

It is time to start regulating this form of AI. And maybe look at some other technologies that have wiped out some jobs (which by now is too late).
This will be great help for lots of things. We're looking into it in our company too.

As for school if one is still keen on essays make them write in class by hand. Most school essays by most students are just busy work anyway.
Owendonovan United States of America
Posts: 1002
Joined: Wed Dec 08, 2021 3:08 am
Location: NYC
Has thanked: 941 times
Been thanked: 713 times

Re: Science/Techno Babble Random, Random

#542

Post by Owendonovan »

ponchi101 wrote: Thu Dec 22, 2022 12:32 am I understand that Climate Change is one very serious threat to humanity. But not far behind (and I do not know if they are behind) stand Work Automation and Artificial Intelligence.
Here is one example:
AI can now write like a human. Some teachers are worried.

The Gist: A company called OPENAI has released a language model that is so complete that it can produce essays that are impossible to tell from human-produced ones. Meaning that, in a short while, every idiot in school could deliver an essay that the teacher could not tell it would be produced by AI.
Imagine the consequences for any and all writers: newspeople, scripts, novelists, anybody making a living by writing.

It is time to start regulating this form of AI. And maybe look at some other technologies that have wiped out some jobs (which by now is too late).
Who will decide what is AI once AI is fine tuned, people or the AI?
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#543

Post by Suliso »

Some philosophy professor was asked whether he's not worried about AI writing essays for his undergrads. His answer: Not at all. Students will write their essays with AI and then send them over to my grading AI. With formalities out of the way we all will be free to learn whatever we want.
User avatar
ponchi101 Venezuela
Site Admin
Posts: 14722
Joined: Mon Dec 07, 2020 4:40 pm
Location: New Macondo
Has thanked: 3857 times
Been thanked: 5567 times
Contact:

Re: Science/Techno Babble Random, Random

#544

Post by ponchi101 »

Suliso wrote: Thu Dec 22, 2022 2:47 pm Some philosophy professor was asked whether he's not worried about AI writing essays for his undergrads. His answer: Not at all. Students will write their essays with AI and then send them over to my grading AI. With formalities out of the way we all will be free to learn whatever we want.
I know you run on the optimistic side (as opposed to me) but, you really don't see any problems with this?
The software is not meant to HELP you write an essay; it is meant to write the essay for you. I doubt that there can be any learning in clicking a button and getting, almost instantaneously, an essay on, let's say, William Blake's literary output.
The professor seems to be also sitting very comfortably on his (probably tenured) laurels. So fine, his AI grades their AI. How long before his AI replaces him? Very tongue in cheek, but as I said before, how about people that write for a living? We have at least one member here that does that: Megan. How fast would her publication be able to replace her if they can get GOOGLE-WRITER? Or, for that matter, how long would it take for an entire magazine to be "published" by just one person, as the sole thing would be for such a person to plug in the "requests" for articles to the AI and let it spit out all of them?
This next comment is personal but please take it as a true question. It seems to me that at times, people can't see that something innovative will affect negatively a very large group of people. For example, in my industry, almost the entire surveying sub-section has been wiped out because GPS systems have become so accurate and precise that you need only 2-3 surveyors to process the data. In the past, we needed 20-25 surveyors per group, plus their assisting crew, to map a project. They all are now gone, with no other options (and it was a highly technical career). So, the question: does it not matter in any way that an entire profession could be wiped out because some AI is implemented? It has happened already: travel agents, surveyors, drafts-people and graphic designers, and, of course, countless lesser-skilled activities.
Does it not matter at all that these AI takes away from the pool of jobs available?
Serious questions, all of them. Would love to hear you out.
Ego figere omnia et scio supellectilem
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#545

Post by Suliso »

I assume the professor didn't mean that students would learn anything by letting an AI write the essay. Instead he meant that these mandatory essays have little value in the first place. Whoever wants to learn something will do so by him/herself.

My argument is not actually that it's good that all those people lost their jobs. It's more that it's useless to try to stop technical progress. Short of some very specific niche cases (nuclear bombs, chemical weapons etc) it has never ended well for those who try.

People who write generic opinion pieces will likely soon be out of a job. Those who actually go out, interview people, discover something new will stay on much longer. AI will not fly to Florida to interview latest hurricane victims.

In one case at least I'm glad of the change. I travel a lot and would be very annoyed if I had to deal with any travel agents.

Right now I'd advise against going into any profession which can be done entirely with your computer with no physical component of any kind. Painting good, graphic design stay out. Unless you can be a very high level AI researcher.

Btw the current version is not really an AI. It only amalgamates already existing information and makes derivative versions. True general purpose AI might appear, but it will still take some time.

Isn't it ironic that computers find it way easier to write good articles than reliably drive a car. The latter endeavour seems to have stalled despite the enormous funding devoted to it.
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#546

Post by Suliso »

Fully online education is already possible, no AI needed. It is however an inferior form of learning for a myriad of reasons. Particularly for children and teenagers.
User avatar
ponchi101 Venezuela
Site Admin
Posts: 14722
Joined: Mon Dec 07, 2020 4:40 pm
Location: New Macondo
Has thanked: 3857 times
Been thanked: 5567 times
Contact:

Re: Science/Techno Babble Random, Random

#547

Post by ponchi101 »

Suliso wrote: Thu Dec 22, 2022 4:39 pm I assume the professor didn't mean that students would learn anything by letting an AI write the essay. Instead he meant that these mandatory essays have little value in the first place. Whoever wants to learn something will do so by him/herself.

...

People who write generic opinion pieces will likely soon be out of a job. Those who actually go out, interview people, discover something new will stay on much longer. AI will not fly to Florida to interview latest hurricane victims.

In one case at least I'm glad of the change. I travel a lot and would be very annoyed if I had to deal with any travel agents.

Right now I'd advise against going into any profession which can be done entirely with your computer with no physical component of any kind. Painting good, graphic design stay out. Unless you can be a very high level AI researcher.

Btw the current version is not really an AI. It only amalgamates already existing information and makes derivative versions. True general purpose AI might appear, but it will still take some time.

Isn't it ironic that computers find it way easier to write good articles than reliably drive a car. The latter endeavour seems to have stalled despite the enormous funding devoted to it.
My further questions (in order from the BOLD):
Your faith in human nature far exceeds mine. The believe that newer generations universally have the drive to learn new things just like that is something I don't share. I have a collection of nephews/nieces that are simply un-interested in so many things I have my doubts.

AI does NOT have to go to hurricane area. Coupled with the DEEP FAKE technology, how much longer before an AI/Deep Fake reporter calls somebody over a cell phone and interviews that person? Even better, because the AI could record the interview and immediately produce the article. In the future? Sure, it is not here yet. But that essay-writing-AI seems to be on its way.

I had such a great travel agent we are actually close friends. She helped me so many times... ;)

Why are you expecting general purpose AI, and then and only then, worry about it? How about if tomorrow somebody generates Chem-AI, an AI software that analyzes chemistry problems? You are in the business of THINKING; if that were to happen, would you be replaced? Why do you think that only when we get GENERAL AI should we worry? How about SPECIFIC AI? Civil Engineering AI? Or MATH-AI (like THIS ONE)? It is not only people in some skill-sets level. Once AI reaches you, you will be replaced. How will that feel?

The self driving problem is simply compounded by US, random-humans. Think about it: if you get rid of ALL humans driving on the road, FSD cars would have it much easier, and the algorithm would work faster.
Ego figere omnia et scio supellectilem
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#548

Post by Suliso »

We can certainly worry, never too early for that. :)

In the distant future all kinds of things are possible. I think we're still far from AI showing any initiative to do something. In the realm of humans not everything can be done by phone. Actually even worse - hardly anyone myself included answer a phone from an unfamiliar number. Someone will have to go out.

As for chemistry AI people are trying hard to make one, including in our company via collaboration with IT folks. There are a couple of stumbling blocks. First of all AI of this sort doesn't invent anything - it only analyzes information out there already. Chemistry is an experimental science, new discoveries are not made by thinking alone. Ok, you might say 90% of cases require no invention. You'd be right, but there are some other things developers struggle with. AI needs a training set, but the chemical literature is incredibly biased towards positive results. AI training ideally needs equal amount of negative data and it doesn't exist. There are some more technical difficulties, but the two above are the easiest to explain.

Autonomous driving is very difficult because it's in an unpredictable environment with many edge cases and making serious mistakes have unacceptable consequences. Having only AI driven cars would help a lot, but not entirely. There are still pedestrians, animals, construction work etc...

I think I'm reasonably safe for the 15-20 years left in my working life. Maybe I live long enough to see all science and engineering being done by computers alone, but I kind of doubt it.
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#549

Post by Suliso »

I played around with ChatGPT. Very convincing answers, BUT quite a few are dead wrong. Especially when asked about Latvia specific topics (in both Latvian and English) to which Google easily finds the right answer.
User avatar
ti-amie United States of America
Posts: 22996
Joined: Wed Dec 09, 2020 4:44 pm
Location: The Boogie Down, NY
Has thanked: 5304 times
Been thanked: 3292 times

Honorary_medal

Re: Science/Techno Babble Random, Random

#550

Post by ti-amie »

“Do not grow old, no matter how long you live. Never cease to stand like curious children before the Great Mystery into which we were born.” Albert Einstein
User avatar
Deuce Canada
Posts: 4531
Joined: Wed Dec 09, 2020 5:52 am
Location: An unparallel universe
Has thanked: 336 times
Been thanked: 977 times

Re: Science/Techno Babble Random, Random

#551

Post by Deuce »

ti-amie wrote: Mon Dec 26, 2022 8:01 pm
Conclusion: ALL social media is crap.
R.I.P. Amal...

“The opposite of courage is not cowardice - it’s conformity. Even a dead fish can go with the flow.”- Jim Hightower
User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#552

Post by Suliso »

User avatar
Suliso Latvia
Posts: 4405
Joined: Fri Dec 11, 2020 2:30 pm
Location: Basel, Switzerland
Has thanked: 274 times
Been thanked: 1453 times

Re: Science/Techno Babble Random, Random

#553

Post by Suliso »

People have found a way to really fool the current version of ChatGPT. Turns out it's grammatically correct and can play with language at undergrad level, but doesn't possess a mathematical logic of a 12 year old...
User avatar
ponchi101 Venezuela
Site Admin
Posts: 14722
Joined: Mon Dec 07, 2020 4:40 pm
Location: New Macondo
Has thanked: 3857 times
Been thanked: 5567 times
Contact:

Re: Science/Techno Babble Random, Random

#554

Post by ponchi101 »

Ok, great.
But I wonder how many people would flunk the same question. I am pretty sure the answer would not be zero. ;)
Ego figere omnia et scio supellectilem
Owendonovan United States of America
Posts: 1002
Joined: Wed Dec 08, 2021 3:08 am
Location: NYC
Has thanked: 941 times
Been thanked: 713 times

Re: Science/Techno Babble Random, Random

#555

Post by Owendonovan »

This telescope is gonna give us amazing photos for at least a couple decades. Stunning.
https://www.nytimes.com/2022/12/27/scie ... scope.html
Post Reply

Who is online

Users browsing this forum: No registered users and 6 guests