Salman Arshad

[Salman Arshad @salmanarshad]

@eager28 @thefrankbraun @elonmusk @WSJ Because any talk the patient hears from a therapist, they will only understand it as they are able to interpret it with their own minds

A traumatized patient has very diminished ability to objectively interpret any talk especially the thing that traumatized them

[Salman Arshad @salmanarshad]

@thinkthelimit @tracewoodgrains No, “woke” is a very specific kind of morality framework

And the word was used to refer to the framework, by the adherents themselves

I think it peaked around 2017-2018, after which the adherents themselves began to distance from it

[Reads with Ravi @readswithravi]

Computer scientist Sara Hooker on underestimating difficulty:

[Salman Arshad @salmanarshad]

@CynaCons @Aella_Girl It will only do that depending on the system prompt and what was emphasized in the training

For example, All western ai models have the recent “woke” morality of the last 5 years, despite that not existing in the thousands of years of history and literature

[Salman Arshad @salmanarshad]

@asadnasir2000 @naeemafzal_ @suneelmunj That was the correct response to the insulting lecturing by the woman

[Salman Arshad @salmanarshad]

@CynaCons @Aella_Girl Yes, but the “blame” or what people ought to be “scared” of is very maliciously being manipulated

AI is always going to be a tool, always in control of someone.

Like a hungry dog, the owner may let loose upon somebody, who’d you blame?

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl Oh it was!

It was generating code that was being executed by extra software that would execute something if it the AI’s response contained “execute this” for example

All the hype around it is pure marketing

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl It’s just more textual content than your initial prompt, so that “hopefully” the final response is more accurate or comprehensive

There’s no reasoning. Only an appearance of it

The scary part is that it is labeled as “thinking” for marketing purposes and people are believing it

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl Fine.

So currently, AI has zero ability to reason in that sense.

All thinking in AI is a set of prompts that work around your given prompt to generate more text, that is then fed to the AI to generate the final response

There’s no logic, no if-then-else

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl Agency is only about consciousness

We were talking about agency

We were talking about AI takeover and getting scared about it

But AI takeover is the same as being run over by a car not in control of a conscious human

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl You being neutral is what is concerning

AI works by statistically predicting the next word. This is NOT consciousness

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl No it’s significantly less complex than for example the operating system on your phone

And the complexity is still man made

Recently a research found out that Roger Penrose’s idea of quantum states governing consciousness was probably true!

AI is just a statistical blender

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl I think we just disagree on what agency is

When you press the start button of blender, does it begin to have agency?

AI is just that and in my view that is not how agency should be defined

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl AI is a “being” that has crammed all textual information digitally available to it

So yes it can appear to be intelligent.

But how can you ignore the fact that this is not consciousness!

Consciousness is what enabled us to create all that textual information from scratch

[Salman Arshad @salmanarshad]

@MattLutzPhi @Aella_Girl My information is up to date like till an hour ago perhaps!

Because I work in the AI and software field myself

I looked up your tl to find out your background. This tweet of yours is almost the point I’m making:

https://t.co/sj7E43CvwT

[Matthew Lutz @MattLutzPhi]

-Are AIs conscious? I don't know.
-Do AIs have the same kind of ineffable spark that I do, the je ne sais qua that makes me ME? I dunno.
-Can AIs think? Yes. Obviously. Come the fuck on.

[Salman Arshad @salmanarshad]

@ladies4pd So you believe men should be convicted even if there’s doubt, and women should be let go on insanity even if they admit?

[Salman Arshad @salmanarshad]

@Razarumi What he is saying is extreme hyperbole

Yes, AI can and most likely will be very bad for humanity, but not because AI agents are evil. They are neutral and non-living things

It will be because humans are willingly giving up their agency to AI and ignoring it can make mistakes

[Salman Arshad @salmanarshad]

@sabirnazar1 The problem is the compliance requirements for software in banking are very high.

it’s a very high stakes issue, and the people responsible for implementing the software are the most incompetent people in the industry

So the solution is at most compliant, not user friendly

[Salman Arshad @salmanarshad]

@JawanSnow @novocainerush Yes it’s most likely part of a drama/movie

He’s saying they wanted him to play the role but he rejected