[lbussell]

If I had to guess, we’re less than 5 years away from seeing real-life C-3PO.

[seanr88]

I would disagree. All of what we are seeing from this latest surge in AI is essentially jumped up predictive text. To get to C-3P0 there is a whole additional layer of Intelligence needed. C-3P0 can make plans and execute those plans. This latest wave cannot reason about the world, it does not know or understand the world it just assembles words (and here motions) in a way that we value. It is not planning anything.

[modeless]

That's the easy part. Making high level plans is trivial compared to the fine motor control and dexterity and sensing necessary to do things like turn a T-shirt inside out or install a fitted sheet or crack an egg or whatever. If you give me a robot with all the fine motor skills necessary for all the steps to cook a meal but no planning capability whatsoever, I'll have that robot cooking your dinner within a year.

[epolanski]

C3PO is a translator droid that is basically stupid at everything else (other than math or facts listing, as he's a robot).

So yes, I think he seems a reasonable target.

[moffkalast]

I think you're giving C3PO too much credit for the bumbling idiot he usually is when on screen. Well aside from calculating the odds of successfully navigating an asteroid field, but I'm sure GPT 4 will let you know what that is just as easily, as well as translating any language into any other language which is supposedly 3PO's whole schtick.

Also:

> can make plans and execute those plans

https://github.com/antony0596/auto-gpt

> cannot reason about the world, it does not know or understand the world it just assembles words

They can reason a surprising amount given that they only work with text. With vision/actuation encoding there's potential for far more. Remember, it doesn't have to be smart or conscious as long as it gets the job done with cold hard statistics while just appearing as such. A submarine does not swim but crosses the ocean just the same.

[edgarvaldes]

Considering that C-3PO waas verbally very competent, but somewhat clumsy, I agree. Even the "old" Boston Dynamic robots are more agile than 3PO.

[darkerside]

Hmm, perhaps that clumsiness was a programmed affect designed to disarm those he interacted with. After all, as a translator and master of protocol, outshining his master in any way would be quite against his programming.

[nradov]

As long as you don't mind the extension cord.

[modeless]

I'll step over an extension cord all day long if it's powering a robot that does the laundry and the dishes and the house cleaning for me.

[jononomo]

If I had to guess, we're over 50 years away from seeing a real-life C-3PO -- hopefully before I die. I think self-driving cars are even further away than that, however.