Main menu

Pages

Science Uprising 10: Ask rude questions about AI

featured image

In Episode 10 of Science Uprising (September 21, 2022 at 10:35), we’ll see why computers don’t rule despite ultra-fashionable TED Talk-style apocalypse claims.

The short film starts with the robot Sophia, who hopes to play a big role in managing the health of the elderly.

“Hello world.” (0:13)

“What emotions do you feel when you are awake in life?”

“strange.”

great. (Wow…!)

The film then cuts to Nick Bostrom of the Oxford Institute for Future Anthropology, announcing to a heated gathering: Superintelligence with such technological maturity would be very powerful, and in at least some scenarios, machines would be better inventors than they are now because they would get what they needed. “

Various celebrities have offered similarly dire prophecies, now very well known.

Will the AI’s success in Jeopardy or Go carry over for the rest of its life? After all, these games involve narrow, well-defined tasks. When AI is given free-hand tasks, things quickly get chaotic. (2:13).

Yes, it is. That’s why IBM’s Watson failed in medicine.

Robert J. Marks Professor of Computer Engineering, you, “The challenges of AI are more than the media will allow. The main limitation of artificial intelligence, in a nutshell, is the algorithm.

“A chocolate cake recipe is a recipe and an algorithm. Inputs are all the ingredients and there are instructions. How to preheat the oven, how long to cook the cake, how to apply the icing…” (2: 57)

Then we hear another question that is rarely or directly asked in the world of TED talks. Can instruction sets, algorithms somehow replicate the human mind?

Oxford mathematician John Lennox said: Roger Penrose is very interesting on this topic. they are not computable. “

Cut to Dr. Marks: “There are things that have been proven to be non-algorithmic. If something is non-algorithmic, it means it is uncomputable. It’s possible, and it’s not speculation or wishful thinking, it’s a proven fact, and it turns out that this non-algorithmic aspect translates into human capabilities, creativity, senses, and comprehension. Speaking of non-algorithmic properties, there are some that are not reducible to computer programs (4:00).

Dr. Lennox says again: It is not true intelligence.
Computers don’t think, they don’t care. “

Oh yes, but the narrator asks another question. What about these talking robots? Some of them look very lifelike.

3D rendering of a female robot looking sad.

In fact, the robot Sophia flashes onto the screen again, giving off the “meaningful” mundane words provided for that program. “What do you think about humans?” “I love my fellow humans. Wow!

When humans say this crap, we know we are flubbing And so is everyone else. Hearing it from something where sincerity hardly matters…is a little painful at best. The fun ends when Sophia is asked a question with no pre-recorded response.

Selmer Bringsjord, director of the AI ​​and Reasoning lab at the Rensselaer Polytechnic Institute, talks about the Lovelace test he developed. The test was developed to determine whether AI is “creative or simply combines and matches works of human creativity according to pre-programmed algorithms.” ” in short, creator The machine operator must consider the originality and consistency of the machine’s work. (6:06) It’s not happening.

Note: The Lovelace Test is named after computer pioneer Ada Lovelace (1815–1852). Ada Lovelace stumbled upon the work of Alan Turing (1912–1954) more than a century later and “wowed” him by finding that his Turing-She-Machine approach lacked something essential.

For example, as Dr. Marks points out, AI can combine all of Bach’s works to produce a Bach-like sound. But that’s not where Stravinsky comes from. Of course, I also add Andrew Lloyd Webber. It is by its very nature unthinkable out of the box.

Philosopher Jay Richards makes the distinction: They work at the level of operational rules. An agent — an intelligent person — works at the semantic level, the semantic level, to understand what a symbol means and what it stands for. Machines are not. they’re just manipulating them. Software is software. (7:59)

That’s true, but as we brace ourselves for all the TED talks we’ll hear in the next few years, the big question arises. (8:20)

The masked narrator of the film states, “The belief that machines will become alive and conscious is unfounded.
Scientifically, it is based on materialism. Materialists argue that since man is only a machine, he should be able to replicate himself with a machine. But the proof? “

When you come up with an answer that isn’t in the script, someone might say: Only those who are not hypothetical seek proof. In this case, the façade collapses when the evidence is called for.

Then we face the reality of being unique as human beings.

Comments