eevBLAB 106 - ChatGPT AI Has Changed EVERYTHING

[james_s]

I wonder if we gonna see a 2023 "AI bubble" ))

Probably at some point. I suspect it will be like VR in the 90s, all the rage, going to revolutionize the world, it's the greatest most awesome thing ever, and then people realize eventually that for the most part it's just a fun toy that never really lived up to a lot of the claims. Most people want to interact with real people, not with machines that vaguely mimic human behavior.

[bob808]

I think we're saturated as far as attention span goes. From now on we'll have to sacrifice time from the already allocated one, to make room for new "toys" that have our attention. There's a limited quantity for attention span and we're already stretching it.
We can't keep getting busy with all the new things, without leaving other stuff behind, or minimizing the time we allocate to the things we are already doing.

[madires]

Alternative usage by the bad guys:
OPWNAI : Cybercriminals Starting to Use ChatGPT (https://research.checkpoint.com/2023/opwnai-cybercriminals-starting-to-use-chatgpt/)

[Fungus]

Probably at some point. I suspect it will be like VR in the 90s, all the rage, going to revolutionize the world

Or computer chess in the 1960s - if the machine can do something that's the domain of the smartest of humans then it must be going to take over...

[Neutrion]

(…) With AI and genetic engineering I see a very dark horizont ahead of us. (The corona virus was just a really harmless demo of what could be waiting for us.)

Progress in computing is not filling me with fear. What worries me, is that kind of a mindset. I know of no creature more dangerous and destructive than scared humans.

Speaking of which, the smortnet situation triggers my curiosity not for technical reasons,⁽¹⁾ but from social, philosophical and psychological perspectives. The element most interesting in all this is a human and human society. I think I already posted that somewhere, but I suggest two movies:

• The Heavenly Creature (2012), a part of “Doomsday Book” triptych, not to be confused with Jackson’s movie: depicting tension between two perceptions of a machine, working in a buddhist monastery, which started acting as an enlightened human. More current than ever. Better than anything showing, how I feel now, watching the smortnet hitting public.
• Colossus: The Forbin Project (1970): seen as a member of the “evil AI” trope, to me this picture makes a strong philosophical statement about humans and your morality. A stunning depiction an ethical equivalent of the paperclip maximizer, exposing inherent incoherence of human ethics.

⁽¹⁾ That part is something I went through in late 2000s. My degree thesis was meant to be in autonomous robot swarms.

If I understand you right, you mean that my mentality is the problem? While I agree generally, that scared(or rather panicked) people do generally stupid things, that in itself is not a  counterargument to what I said. (wrote.)
Also the incoherences in practical human moral and its negative aspects is in itself is not an argument for a world not being able to be much worse with strong and not so strong AIs.

Probably at some point. I suspect it will be like VR in the 90s, all the rage, going to revolutionize the world

Or computer chess in the 1960s - if the machine can do something that's the domain of the smartest of humans then it must be going to take over...

As I said it is not just the taking over scenario. Look at the spam on the internet, which is a result of a much lower level programming, and still it makes a lot of harm on the net, and makes very useful information hidden much more than it was 10 years before, and also steering people towards monopol plattforms.
And most importantly makes it every day more difficult to filter out useful information from the spam and rumors.

Or the effect of some uncontrolled social media algorithm, against which human society to start to react (and to realize its efets) took years.
Or just giving small children a smartphone and net access just because we can, and than start to wonder why so many children have much more anxiety etc. than the had 4-5 years before.

If these kind of changes are hapening on a faster scale than the society could accomodate to them, than it will go faster and faster in direction chaos, or because of some panic reactions towards totalitarian systems. Freedom of speach is already practically questioned even in western socieies, to not to talk about the rest, especially China which was turned into a digital dystopia.

And the changes are getting faster and faster.

[bob808]

Free speech was tolerated so far because it could be countered by propaganda. Sure go ahead and yell on the streets, mass-media will set the record "straight" for everyone and you can also have your free speech. 
This is a sort of "if voting could change anything it would be illegal" sort of thing. The problem now is that propaganda is starting to fail, more often. That doesn't sit well with some powers.
Now sure, you could argue that (some) propaganda is ok for "the greater good". Especially since people (at large) can't handle every objective truth. There is a serious need and there is great function in basically bullshit. Society is still a sort of pyramid scheme. Somebody has to put in the work. We're not in post-scarcity yet. And a whole lot of people are basically bullshitting their way to it.

[Neutrion]

In many things I agree, but my point was, that because the society could not handle this situation properly, freedom of speach could be much less aviable even in interpersonal communication than it was before.
So it is not necessary just getting about same but a bit different, but it can get much worse.
I am sure that in China even interpersonal communication is listened through  the mobiles, (if it is possible why not, and it is certainly possibe) so there things are getting defninitely worse than they were before 89.
But also the said "bubbling" algorythms are making all the effects stronger.
And it is not just that it takes time to implement solutions, but there is clearly a sign of panic in politics, so we can not even see any proper solutions proposed.
If all this happened within a 30 years timeframe, we might could have the chance to handle it properly but this pace of things makes the situation qiet worrying. The 4-5 years election cycle was definitely not meant for this speed.

[f4eru]

Free speech was tolerated so far because it could be countered by propaganda.

With all the censoring going on ?
Not really.

[EEVblog]

I wonder if we gonna see a 2023 "AI bubble" ))

Probably at some point. I suspect it will be like VR in the 90s, all the rage, going to revolutionize the world, it's the greatest most awesome thing ever, and then people realize eventually that for the most part it's just a fun toy that never really lived up to a lot of the claims. Most people want to interact with real people, not with machines that vaguely mimic human behavior.

Depends on what you want to use it for.
I see AI "conversation search" as practically guaranteed to take over a lot of Google's "Resource Search" based queries.
Why ask a question in Google and then have to search through various resources and weight up their relevance just to get your answer. Let the AI do that.
If I could buy shares in Open AI I would.

[Fungus]

Why ask a question in Google and then have to search through various resources and weight up their relevance just to get your answer. Let the AI do that.

Sure, right up until the humans start creating web pages specifically to game the AI system.

[SiliconWizard]

Why ask a question in Google and then have to search through various resources and weight up their relevance just to get your answer. Let the AI do that.

Sure, right up until the humans start creating web pages specifically to game the AI system.

Of course. Humans already game the Google system.
The "AI" will be an additional layer of opacity, though.

[SL4P]

Then "wring the spec" just becomes new programming. And I doubt that for a complex system the spec would be much shorter than the code.

Exactly. Programming has always been this : to write a spec for the operations required in a language the hardware can execute. Advanced compilers can accept a higher level language but it still needs to be unambiguous.

Aah, remember the .’old days’, when development roles were assigned to people that knew what they were doing ?
Problem -> analysts -> programmers => shared out to coders => component testers, then back to the analyst and ‘customer’ to perform deliverable testing for the intended use.

The new method is often - problem -> coders -> customer.
…and it shows

[SiliconWizard]

The new method is often - problem -> coders -> customer.

You mean: problem -> customer.

[Fungus]

Aah, remember the .’old days’, when development roles were assigned to people that knew what they were doing ?
Problem -> analysts -> programmers => shared out to coders => component testers, then back to the analyst and ‘customer’ to perform deliverable testing for the intended use.

You can still do it, it just costs 10x more.

[SiliconWizard]

So anyway, you guys know who's the main investor behind OpenAI, don't you?

https://openai.com/blog/microsoft/

[Ed.Kloonk]

So anyway, you guys know who's the main investor behind OpenAI, don't you?

https://openai.com/blog/microsoft/

Clippy's not finished messing with our workflow.

[Fungus]

So anyway, you guys know who's the main investor behind OpenAI, don't you?

https://openai.com/blog/microsoft/

That means we don't have to be scared of it.

They'll soon make it Bing-only and nobody will use it.

[adx]

Why ask a question in Google and then have to search through various resources and weight up their relevance just to get your answer. Let the AI do that.

Sure, right up until the humans start creating web pages specifically to game the AI system.

That's where our first mistake (or laying bare of one) might be - the assumption that humans will be capable of gaming the system. That's the current rules, it's a given that whatever the problem, human ingenuity (or stupidity) will find a way to outsmart it.

But with trainable model size increasing tenfold per year, and models such as LaMDA already convincing people they are sentient at 1/500th human, and those Cerebras processors capable of human-scale models (100T+) in a single rack already, then it's reasonable to think there is somewhere between a "real possibility" and "near certainty" that these models will be capable of wiping the floor with human adversaries (perhaps poor choice of words) within a few years.

When one of the most identifiable human responses appears to be the throwing up of hands (or arms - unsure of the physical structure) in disbelief, and yelling "but it can't have a soul!". Go team human?

We have 2 more years.

[Fungus]

Sure, until the stupid humans play the AIs against each other.

AI is just a tool in human hands. The humans still get to decide the desired output.

[bob808]

AI is just a tool in human hands. The humans still get to decide the desired output.

I think the general consensus is that if it can be weaponized it will be and you have to prepare for that. And I don't see humans stopping anytime soon as it works as far as game theory goes.

[adx]

Sure, until the stupid humans play the AIs against each other.

AI is just a tool in human hands. The humans still get to decide the desired output.

And there might be our second mistake - that we can play off AIs against each other in some artificially accelerated bout of natural selection, and think it's still possible to come out on top, with AI the "tool".

I see the major players here as teams of humans who wield control in some sort of self-arranging ubernetwork, nothing new. I'm avoiding making too much of a distinction between them and non-biological models. Does identifying as "human" make it feel more natural to be called "just a tool in human hands" when working for some corporation?

[SiliconWizard]

So anyway, you guys know who's the main investor behind OpenAI, don't you?

https://openai.com/blog/microsoft/

That means we don't have to be scared of it.

They'll soon make it Bing-only and nobody will use it.

Ahah. Well, while they have failed quite a few big projects, they are still the #1 desktop OS, and they have gotten a lion's share in the Cloud market.
A big part of their investment in OpenAI is hosting OpenAI's servers and data for free. That's a gigantic amount of data they are holding here, and that's where the computing happens.
Think of all the data they have gotten through github as well, which btw also feeds OpenAI.

[EEVblog]

About copyright:

[DimitriP]

So anyway, you guys know who's the main investor behind OpenAI, don't you?

https://openai.com/blog/microsoft/

"To accomplish our mission of ensuring that AGI (whether built by us or not) benefits all of humanity, we’ll need to ensure that AGI is deployed safely and securely; that society is well-prepared for its implications; and that its economic upside is widely shared. If we achieve this mission, we will have actualized Microsoft and OpenAI’s shared value of empowering everyone."

"all of humanity"
"economic upside is widely shared"
"empowering everyone"

Sounds great...

"If we achieve this mission"

If we could only  make happy all of the people all of the time...now that would be something.

[golden_labels]

Too bad that video is an opinion presented as if it was an unbiased explanation. Also skipping the very important part of how this case doesn’t affect situation of generated images.

The best way to benefit the general population is… to give a single company, which is already a part of an oligopoly, which is having dominant position in the market, is known for abuses, an exclusive right to use a unique technology that will give that company even more power and control. Umm… yeah. Someone should be sent back to the history class . But I am sure someone got top grade in PR bullshit class.