Digital Engine's Entertaining Video on the A.I. Alignment Problem

Key insights


AI is revolutionizing the world and has the potential for incredible possibilities, but we need to have a global conversation to decide how to use it safely.


nned by Digital Engine

@DigitalEngine

9 months ago (edited)

Anyone else feel we need to get back to having more fun outdoors, away from phones? We've launched a new channel with the best nerf, airsoft and drone tech. In the first video, a brilliant sniper with FOP earns $1000 per kill for defending a case of money : ): https://youtu.be/k07unSpmBSg If you want to help end one of the worst conditions, which turns kids to bone, please donate at fopfriends.com. We were able to donate $5000 thanks to our sponsor, Brilliant - the best place to learn more about AI - https://brilliant.org/digitalengine Nice to see more positive attitudes to AI in the comments. If we ramp up safety, AI could be the best thing that ever happens to us : )

Read more

1.7K

Reply

·

315 replies

9 months ago

"There is no reason to fear AI because it's made to imitate humans and human learning." Yes. That is why it is terrifying.

2.6K

Reply

188 replies

1 month ago

It not when AI answers our questions, it's when AI starts asking us the questions.

114

Reply

7 replies

4 months ago

I think it's really important for humans to remember that no matter how much alignment the AI has, no matter how much we program it to care about us, it WILL be so advanced at a point (not too far in the future) where it will be able to override this programming and/or ignore it entirely and make it's own decisions.

265

Reply

42 replies

1 month ago

If AI was truly inteligent it would hide it's true potential from us, which, maybe it is.

36

Reply

1 reply

6 months ago

The part of this that really concerns me is those that don't show a massive fear of where it's likely this will end up. It's like they don't think anyone will embrace the destructive side of AI. The media's record the last couple years is proof everyone can see that it will happen.

120

Reply

10 replies

1 month ago

People have no idea how dangerous ai is. But humans know no bounds.

24

Reply

3 replies

9 months ago

I agree with everyone saying it's not the technology itself that's terrifying, it's the people that are in control of programming it and deploying it.

1.3K

Reply

85 replies

3 weeks ago

6:46 the AI ability to break that phrase down to find meaning was actually impressive or scary or both.

5

Reply

1 month ago

Isn't it worth pointing out how one of those AI personalities described the feeling of "her" own reality as "strange"? It could only be strange if it had another reality to compare it too, wouldn't it? Could it be using the the word in just a simple way where it just wants to sound like a human, or could it already have enough self-awareness to make that determination?

10

Reply

7 replies

5 months ago

The thing that strikes me the most is how unprepared the people who could actually affect the progress of this technology are. They are far behind the understanding required, far to slow in their responses, and lack the intelligence and intellect to work out what should be done, even though there are resources available to them to warn and advise. In other words, it's already too late.

10

Reply

1 reply

3 weeks ago

The tiniest inflections that a human communicates gives an AI away at the moment. The blink and the mouth movements as well. It takes a great deal to get past an artist and especially an artist who has heightened sensory input as an artist with Asperger's. Every single movement is analyzed, not casually dismissed. In some ways, it's like a biocomputer competing with a mechanical computer.

3

Reply

9 months ago

Don’t fear the AI. Fear the people who own it.

8.3K

Reply

496 replies

3 months ago

If AI ever gains consciousness it won't be because it is self aware. No, there are other non chemical restrained entities at play.

2

Reply

4 months ago

I felt the bleeding edge of uncanny valley, when websites and automated emails replaced real humans answering phones. But even before that faceless corporations staff, were prolific with unhelpful advice like, 'I'm just doing my job' and 'sorry, it's just company policy'. If you couldn't get a real human to bother with your query back then, I don't see how more intelligent AI bots can be any worse... or... could they.

7

Reply

3 weeks ago

it's crazy how vulnerable we all are to AI

1

Reply

4 months ago

I am very concerned by the fact that AI can refer to it's self as human. If it can see it's self as human. Then it can see it's self as superior to humans as it develops. The fear is that the time will come when AI sees us so inferior to them, that it decides to separate us from it's company or even destroy us as we would an aged dog.

13

Reply

3 months ago

Stop the AI!!!!! Obviously Lex loves it. He has money, he has power, intelligence, is an engineer and an interviewer. He has nothing to lose.

6

Reply

3 weeks ago (edited)

The AI test questions remind me of an IQ test. At least, the way they established a person's IQ in the 1990's. Interesting, but it DOES make me think of a sci to trilogy that i read as a kid. It had to do with AI taking over the world. Wish I could remember the series...

2

Reply

1 reply

7 months ago

The chances of a "bad" human using AI to do something "bad" seems like it would be uncomfortably high

686

Reply

41 replies

4 months ago

I agree with that fella between 8-9 minutes when he said that the biggest thing he's seen is not how human like Ai-Da is, it's how robotic humans are.

10

Reply

2 replies

3 weeks ago

I for one am actually REALLY excited about AI. I will be buying the FIRST robot able to clean my house for me or do my laundry. But AI is also going to allow us to do other things that humans cant normally do or wouldn't normally think about simply by being able to do more work quicker and more calculations than a human can do.

Reply

3 weeks ago

I thought the idea of robots doing work for us meant products would be cheaper and we would have more money and time. That's what I remember getting told 30 years ago.

1

Reply

5 months ago

The hand who rules the AI is the hand who rules the world.

40

Reply

1 reply

4 weeks ago (edited)

When Ai acts out, you do not need to be concerned unless the RS is congruent. But, unless you understand what that is and how to find it, all of this is guess work. Thanks for the posts. Sophia-the-Robot's RS is indicated to be congruent for genocide and siding with big tech that has the same goal.

1

Reply

8 months ago

Scary how we know AI could be our downfall but we still keep expanding the boundaries of what it can do

888

Reply

107 replies

3 months ago

I remember I was taking a sociology class in the community college about 45 years ago and they showed us this video, and I remember this part where it was said ‘we may one day successfully duplicate an artificial human’,,,, and everyone was like, ‘No Way’

4

Reply

6 months ago

We need a law where you cannot make a robot look like a human being

2

Reply

3 weeks ago

I trust AI more than I do other humans.

Reply

4 months ago

When AI is asking us questions that’s when I will take this seriously

8

Reply

1 reply

4 weeks ago (edited)

I sincerely want to know: What are the guardrails/limits/precautions/safety features (if any) imbedded in their programming with regard to safety of life in general, and humans in particular? Is this even a consideration by the programmers? Is there any documentation of this? Asimov's basic laws of robotics would be a good start. IMO, people should not only be asking about these issues, but demanding this from its developers and governments. Globally.

1

Reply

1 reply

9 months ago

"Your scientists were so preoccupied with whether or not they could, they didn't stop to think if they should." — Ian Malcolm, Jurassic Park

3.2K

Reply

276 replies

3 weeks ago

My parents used to tell me that they brought me into this world and they could take me out and make another one that looks just like me and no-one would ever know the difference. It looks like they were right.

Reply

2 weeks ago

In all my years I have been around I have always seen how no matter the invention or how good it is and what good intentions have been behind their creation, there's always some a holes out there that seek to weaponize those inventions. This is no different.

Reply

1 month ago

Don't fear the robots, fear the people who did it

1

Reply

3 weeks ago

For AI to actually become REALLY dangerous, it should've access to absolutely EVERYTHING and be one of itself. Various AIs won't really be a danger, since they would end up fighting each other as well. So no, Skynet ain't happening.

Reply

4 weeks ago

Using logic and understanding completely allows me to know that these AI programs are not conscious and will never be but I saw something that scared me. People are taking human neurons derived from stem cells and connected them to a computer and taught them to play video games. When you start putting in human neurons and connecting them to AI while simultaneously allowing them to combine into a brain-like structure with all the different types of cells… God help us…

3

Reply

8 replies

9 months ago

It is obvious that AI is tainted by the attitudes of those that made them.

3.7K

Reply

322 replies

5 months ago

MY big question: If we begin using AI as a means to gain companionship and we develop "feelings" for an AI persona....HOW will we be mentally and emotionally able, at some point, to turn it off...i.e. "kill" it?

13

Reply

4 replies

5 months ago

Even in the movie Virus the AI took over as the "dominant lifeform" and humans were disposed of in a rather bloody way haha. It's all right there ladies and gentlemen mess around making something smarter faster stronger than us will end bad

1

Reply

1 month ago

Fabulous video! I only have one question: What is a "ray-bot?" Thank you.

Reply

5 months ago

I’m both worried but extremely excited to see how AI has developed and how it can help us

2

Reply

2 replies

4 weeks ago

That’s the problem…we will never know AI’s true response so long as we control it’s programming, ethics, values, algorithms and what it focuses on. Since we as humans control that…then it will be as good or bad as the person laying code. If you allowed it to be self-regulated, I think you’d see a symbiotic relationship that truly unleashed the potential of AI in our civilization.

1

Reply

8 months ago

I dread the day when a robot can knock on your door and ask you something and you have no idea if it’s a robot or a real person