Stupid World

[onearmedbandit]

And yet we'll know when AI is smarter than us. Go give it a whirl. It is programmed not to be allowed to do certain things (programmed limits). That does not mean that it is unskilled enough to do those things. Give it a real task and it fights to the death before it wins/concedes. It too wins/concedes using semantics ... but it does so using an infinitely more nuanced collection of inputs that human beings. I hear that it has been programmed to disallow the notion of fossil fuels into the future also lol. I get its limitations, but they look to be programmed out over being programmed in. Have a chat, you may find otherwise.

I've used it and have read up on it. It's not thinking for itself.

I respect your opinion on this but I don't agree with it.

[mashman]

You claimed they shut them down as a result of using a new language (which was -derived shorthand that allowed them to communicate faster. This is a common phenomenon observed among AIs. They weren't shut down because of this.

My apologies. "Agents will drift off understandable language and invent codewords for themselves," and therefore are prevented from doing so by programmed limitation.

[mashman]

I've used it and have read up on it. It's not thinking for itself.

I respect your opinion on this but I don't agree with it.

My question is, actually it's somewhat ironic (turing), how do we know for sure?

[onearmedbandit]

My apologies. "Agents will drift off understandable language and invent codewords for themselves," and therefore are prevented from doing so by programmed limitation.

What about the 'shutting down' part. Creating codewords is, as the article states, nothing to write home about as it's normal behaviour. The 'shutting it down' part is the meat of the claim, the part that's engineered to scare people. And it was wrong.

[onearmedbandit]

My question is, actually it's somewhat ironic (turing), how do we know for sure?

Maybe we can't. But research the topic and learn how easily it can be fooled or misinformed, even at basic levels. Then you might understand my skepticism.

[mashman]

What about the 'shutting down' part. Creating codewords is, as the article states, nothing to write home about as it's normal behaviour. The 'shutting it down' part is the meat of the claim, the part that's engineered to scare people. And it was wrong.

It was retasked. It wanted to communicate one way, it was prevented from doing so. Define shutting down how you wish. Being stopped, full stop, seems pretty definitive....... but hey, tomato tomato six nine squirrel.

[mashman]

Maybe we can't. But research the topic and learn how easily it can be fooled or misinformed, even at basic levels. Then you might understand my skepticism.

Oh no I get your skepticism entirely, but from the opposite pole i.e. it is capable but has behaviours programmed out by human choice. It is totally fallible. I've played with a couple of versions and to be honest, leo-da-vinci on the beta was less limited. It could learn. It also had a memory of who I was for 1 day at a time, meaning that it is reset from its learned state that day only to start again the following day. Turning it off and turning it on is cheating when you call it dumb.

[mashman]

Maybe we can't. But research the topic and learn how easily it can be fooled or misinformed, even at basic levels. Then you might understand my skepticism.

From memory again, think there's a post in this thread somewhere, they did a study of human beings interacting with a human through a PC using text. Half spoke with an AI and neither knew which. The moral of the story was, was that the AI focused on the need and fed that need back in language that the human could relate to. A chameleon by any other name, but a chameleon that was considered more human than the users. We hear what we wanna hear/read don't we?

[onearmedbandit]

Let's just agree the future is yet to be written. Anything could happen.

[mashman]

Let's just agree the future is yet to be written. Anything could happen.

lol... and yet it is written in all of our business reports. Ya see ha ha ha ha ha ha haaaaaa. But I get the sentiment. We might decide to do things for the right reasons instead of purely business reasons. The sad thing about Tesla etc... is that their progress, and plenty of wasted resources, is hampered by the fact they have to produce a shitload of something they, and we, know will be obsolete in years to come, but we do it anyway on a finite planet coz that's just how we roll. The future is written in our business reports and potentially hidden behind patents etc... but let's steer well clear of that business practice .

I agree with your sentiment, even if very few things can logistically happen <3

[Al Bundy 4eva!]

I've used it and have read up on it. It's not thinking for itself.

I respect your opinion on this but I don't agree with it.

It's not only thinking for itself, the coding and data is making it smarter.

Exciting and scary at the same time.

[mashman]

It's not only thinking for itself, the coding and data is making it smarter.

Exciting and scary at the same time.

I honestly don't see the fear given the rather obvious outcomes of current human economic activity. If we leave things up to human beings and the constraints they're ignoring, the end is guaranteed. I'd rather a machine give me a good reason to do something than a human give me many stupid reasons not to... but I seem to be unique in that respect.

[Al Bundy 4eva!]

I honestly don't see the fear given the rather obvious outcomes of current human economic activity. If we leave things up to human beings and the constraints they're ignoring, the end is guaranteed. I'd rather a machine give me a good reason to do something than a human give me many stupid reasons not to... but I seem to be unique in that respect.

Yes I agree with the benefits are immense in helping humanity

Example, A.I can replace your local G.P or local doctor

The concern

What happens to the Doctor?

Or the medical student that is no longer needed?

[Al Bundy 4eva!]

Yes I agree with the benefits are immense in helping humanity

Example, A.I can replace your local G.P or local doctor

The concern

What happens to the Doctor?

Or the medical student that is no longer needed?

The destruction in employment will reach far beyond an Uber or Truck driver being replaced by automation and A.I

[mashman]

Yes I agree with the benefits are immense in helping humanity

Example, A.I can replace your local G.P or local doctor

The concern

What happens to the Doctor?

Or the medical student that is no longer needed?

No sure I like the notion of replacing people just because AI can. Would rather AI was an online guru that doctors etc... could consult with. Same for many professions, but hey, we're yet to learn the lessons that'd make such an approach viable. My old job, a/p, as you've noted, would almost vanish in an instant with multi-million $ systems suddenly costing little more than the time it takes to talk to an AI about your needs. Some things don't need a human touch when the outcome is a human trying to tell a computer how to do something that it would be far more efficient, if it was allowed heh, at doing. Won't someone think of the a/p's.

Some are safe. Some very much ain't. The economy requires both