chatGPT and its cousins are the way of the future ...

[Lurkalot2]

I can see some very big corporations going to the wall pretty quick. Advertising, graphic design, branding. Come to think about it, it will deliver better results than google. (I pay google a monthly bill for 'landings' on my website)
Who would have thought the mighty google being challenged.

I don't profess to being particularly computer literate or to really understanding this topic so stand to be corrected. Wouldn't those big corporations be the ones to embrace AI on a larger scale? Dumping all those talented flesh and blood people and replacing them with the smart computers in the same way big business embraced everything from the spinning jenny to the diesel engine. Anything that keeps the money coming in and preferably ever more money.

[Vortex2]

I can see some very big corporations going to the wall pretty quick. Advertising, graphic design, branding. Come to think about it, it will deliver better results than google. (I pay google a monthly bill for 'landings' on my website)
Who would have thought the mighty google being challenged.

I don't profess to being particularly computer literate or to really understanding this topic so stand to be corrected. Wouldn't those big corporations be the ones to embrace AI on a larger scale? Dumping all those talented flesh and blood people and replacing them with the smart computers in the same way big business embraced everything from the spinning jenny to the diesel engine. Anything that keeps the money coming in and preferably ever more money.

Bingo!

The level of denial out there is quite astounding .. but understandable.

Journalists in particular are the main poo-pooers.
(However, one NYT contributor says that this tech has made his job irrelevant)

I have used chatGPT - and its paid for big brother - for all sorts of things.
Its capabilities are astonishing - and it's only a beta.

There are also running AIs in the video, audio, art and other domains .. a multipronged attack.
This is all adding up VERY fast ... our world will be very different in a years or two.
We may not really notice that, however .. in the same way we accept Google as just another utility.

Skilled people can - are - leveraging it to become 10x more productive.
However, that could mean that the less skilled/experience/AI compatible staff are going to have a very hard time in the future.

Trying to find a job if you are a marginal new graduate will soon suck.

The high-fliers of course will always be OK.

Maybe 80% of jobs in some (many!) sectors are at risk - although inertia, social pressure, legal pressure will block many lay-offs, so for a while many staff will be safe doing 'busy work'.

And of course newer versions of such software are on the way ..

[Vortex2]

FROM Reddit : the gptCHAT sub:

The orange line shows the gptCHAT sub membership .. soaring to the heavens ...

I also believe that Open AI is having to block/restrict new signups as their systems are totally overloaded.

[RevdTess]

Skilled people can - are - leveraging it to become 10x more productive.

From the very start of my IT career I always said my best skill was the ability to know how to look things up quickly - i.e. how to use a search engine.
Now it will be how to get the best answers/work out of AI.

I'm also using it a lot for summarising complex issues in a straightforward way without having to read long technical articles.

I don't get the feeling it's 'learning' as it goes along though? More like a frozen encyclopaedia. I assume at some point it'll start scanning the internet like google...

[Vortex2]

More like a frozen encyclopaedia. I assume at some point it'll start scanning the internet like google...

Yes and yes.

It however acts more dynamically than a frozen encyclopaedia.

One 'trick' is that on top of the cold huge neural network, it maintains a small cache per user, so it has a goldfish-like short-term memory for each user's interactions. This allows quasi-sentient bevaiour.

I really need to see if I can understand the user-facing 'front end' of that software.
The interactive aspect of it is intriguing.

GPT-2 and GPT-3 (the basis for chatGPT) have been around a while, so I'm sure there must be stacks of papers and tutorials explaining it all.

[Vortex2]

There are quite a few AI systems out there.

In a couple of years, when they mature a bit and merge, the world will somewhat different.

[dustiswhatweare]

When this thread appeared I was thinking what's all this then and does it matter to me?

Then a couple of days ago I was on Youtube where I follow a few channels, including one called Cool Worlds, which has a cosmological bent. I'm very careful on Youtube, for the simple reason it has a high percentage nutter content. Don't want to go down the LJ route.

So meet Prof David Kipping of Columbia University, who specialises in exo planet research. It's layman friendly. He decided to investigate chatGPT, and let it sit an exam - ChatGPT Takes A College Level Astrophysics Exam.

This is what happened, and he reaches some conclusions. Enjoy, it's interesting and informative.

https://www.youtube.com/watch?v=K0cmmKP ... CoolWorlds

[Vortex2]

Sheesh, it did way better than me .. and I am supposedly a Physicist.

chatGPT currently has arithmetic problems - but these should be fixed in due course - the current version is only a beta.

We are entering a Brave New World.

[Catweazle]

It's worth remembering that the next gen, coming soon, will be hundreds of times more powerful. Also, if the questions has been worded differently, as they would be by an AI trained researcher, the answers might have been different. Example: use appropriate formulae to answer the following question ......... Might have avoided the cases where AI answered wrong but followed with a correct explanation.

[RevdTess]

I'm noticing some people already starting to feel that they've becoming too dependent on ChatGPT for their work and personal lives and are feeling that when they're cut off from ChatGPT they are somehow 'reduced' and 're-burdened' by tasks that they feel no longer effective at doing on their own.

I guess people have always felt this way about new technology, like the shift when search engines arrives and wikipedia and suddenly you could write essays without stepping into a library and write code without reading a manual. Some technological shifts only affect a small number of people, such as printing removing the need for copyist scribes, and computers removing the need for manual typesetting. But some like ChatGPT will affect almost all work, so that it won't be possible to write an acceptable report or quick-enough code unless you're using an AI.

I still think my sermons are safe for the time being. What will be interesting is when I can feed it all my previous sermons and say "Write a sermon on this bible passage in my usual style" and see what it comes up with. At the moment I tend to read several other sermons on the same passage in different styles and then write what I think is relevant for my own church in my own style. ChatGPT could easily do all of this except that it would not understand the context in which its sermon would be read, unless there was some way for me to tell it. E.g. "Please write a sermon on passage x in my style so that the congregation, who are currently experiencing y may come to understand z". And then I tweak and edit as appropriate. But this may still be too simplistic as a prompt or desired outcome. I'd hate for AI input to make sermons become homogenous and bland - oh, wait...

I may feel a twinge of guilt about this but then remember that every hour saved is another hour available to sit with someone to comfort them. If on the other hand I'd never written a sermon by hand but had always done everything with ChatGPT I might feel a bit more anxious about my dependence on it! But honestly, if ChatGPT could create the powerpoint slides for my 15 minute school assemblies (which all follow the same format) I would use it in a shot.

[clv101]

For anyone who has watched a lot of Star Trek, there are some interesting analogues.

Obviously the 'computer' is very similar Alexa+ChatGPT, but actually the whole work flow of the crew is heavily dependent on underlying technology. The medics just wave their tools over the patient for a cure, the Tricorders see/process all environmental data in your hand. Problems are solved just by the crew having high level discussions and the computer/technology doing ALL the work. No problem in Star Trek requires a week of code development or a month of prototypes in a lab for example.

[RevdTess]

For anyone who has watched a lot of Star Trek, there are some interesting analogues.

Obviously the 'computer' is very similar Alexa+ChatGPT, but actually the whole work flow of the crew is heavily dependent on underlying technology. The medics just wave their tools over the patient for a cure, the Tricorders see/process all environmental data in your hand. Problems are solved just by the crew having high level discussions and the computer/technology doing ALL the work. No problem in Star Trek requires a week of code development or a month of prototypes in a lab for example.

That's a good point. Some 'tech fixes' in Star Trek did actually require the engineers to invent some changes to containment fields blah blah invert graviton flow blah blah reroute the plasma conduits blah blah modulate the nacelles blah blah output from main deflector dish etc. And they often did this together while huddled over a table display in Engineering. But the actual changes all seemed to be done by voice command with the occasional shot of Wesley Crusher tapping a few buttons on the display.

I do feel slightly disappointed with Alexa when compared to ChatGPT though. So primitive!

[Vortex2]

I do feel slightly disappointed with Alexa when compared to ChatGPT though. So primitive!

Just wait for V4 coming out soonish. (This is estimated to be maybe 1000x more capable - at least in database size)

And of course progress (?) will plough on, impacting text, audio, visual and eventually the robotic domains.

[Mark]

The academic world should be worried too......?
Type in the bones of what you want an essay about....., go down the pub with your mates and then submit in the morning....

Teachers nervous about ChatGPT cheats:
https://www.linkedin.com/news/story/tea ... s-6137354/

[kenneal - lagger]

.............
Teachers nervous about ChatGPT cheats:
https://www.linkedin.com/news/story/tea ... s-6137354/

That's one of the problems of continuous evaluation over examinations.