Ensign Steve waxes philosophical on the Singularity, a thrad by Ensign Steve - Page 19

Kamilah Hauptmann · #**451** 09-29-2021, 02:20 AM

That’s extra wow for me since I’ve been role playing in a dystopian cyberpunk genre game as a rogue AI nanny with a warped version of the three laws full of manufacturer rootkits. It’s been a wild mental exercise and that video is my jam, man.

JoeP · #**452** 09-29-2021, 08:56 AM

Quote:

Originally Posted by Ensign Steve

Embodied Moxie - YouTube

Quote:

Comments are turned off.

Kamilah Hauptmann · #**453** 10-10-2021, 12:19 AM

Saturday Morning Breakfast Cereal - AI

Ari · #**454** 10-16-2021, 12:50 AM

Be wary when teaching AIs that they actually have your goal in mind, and not just something that looks so similar to your goal it’s not noticeable during training. For example, did you train and AI to solve a maze and find the exit, or did you train it to head towards the green thing because your exit sign is green?

There’s an interesting bit where we get to see the heatmap of what an AI is ‘thinking’ as it tries to solve a platformer game. Showing things it thinks are good and which are bad, the goal being to collect a coin. During training the coin was always at the end, and shinned brightly in the heatmap as a positive thing, so all good! But during deployment, they randomized where the coin was, and the AI would skip over it to reach the end of the level. The researchers don’t know exactly where the issue is occurring but it would appear the AI learned to first go right until it can’t go right anymore and *then* look for a coin, instead of always be on the lookout for a coin, oh and also go right if one can’t be seen.

We Were Right! Real Inner Misalignment - YouTube

Kamilah Hauptmann · #**455** 10-21-2021, 07:58 AM

Singularity, one step closer.

JoeP · #**456** 10-25-2021, 08:25 PM

I'm mostly posting this news because of a mildly amusing line on Have I Got News For You.

An AI robot artist named Ai-Da went to, um, was sent to Giza in Egypt to host a show of her work.

But Egyptian border guards arrested, um, detained her on the basis of having cameras and a modem.

Egypt detains artist robot Ai-Da before historic pyramid show | Egypt | The Guardian

"It's not known what they are charging her with. A USB cable probably."

ceptimus · #**457** 10-26-2021, 12:50 PM

Quote:

Originally Posted by Ari

Be wary when teaching AIs that they actually have your goal in mind, and not just something that looks so similar to your goal it’s not noticeable during training.

This reminds of part of an essay by Richard P. Feynman on Cargo Cult Science.

There's a lot of other stuff in the essay, but the relevant part here is about some maze rat running experiments carried out by a Mr Young, in 1937.

Quote:

Originally Posted by Richard P. Feynman

For example, there have been many experiments running rats through all kinds of mazes, and so on—with little clear result. But in 1937 a man named Young did a very interesting one. He had a long corridor with doors all along one side where the rats came in, and doors along the other side where the food was. He wanted to see if he could train the rats to go in at the third door down from wherever he started them off. No. The rats went immediately to the door where the food had been the time before.

The question was, how did the rats know, because the corridor was so beautifully built and so uniform, that this was the same door as before? Obviously there was something about the door that was different from the other doors. So he painted the doors very carefully, arranging the textures on the faces of the doors exactly the same. Still the rats could tell. Then he thought maybe the rats were smelling the food, so he used chemicals to change the smell after each run. Still the rats could tell. Then he realized the rats might be able to tell by seeing the lights and the arrangement in the laboratory like any commonsense person. So he covered the corridor, and, still the rats could tell.

He finally found that they could tell by the way the floor sounded when they ran over it. And he could only fix that by putting his corridor in sand. So he covered one after another of all possible clues and finally was able to fool the rats so that they had to learn to go in the third door. If he relaxed any of his conditions, the rats could tell.

Now, from a scientific standpoint, that is an A‑Number‑1 experiment. That is the experiment that makes rat‑running experiments sensible, because it uncovers the clues that the rat is really using—not what you think it’s using. And that is the experiment that tells exactly what conditions you have to use in order to be careful and control everything in an experiment with rat‑running.

I looked into the subsequent history of this research. The subsequent experiment, and the one after that, never referred to Mr. Young. They never used any of his criteria of putting the corridor on sand, or being very careful. They just went right on running rats in the same old way, and paid no attention to the great discoveries of Mr. Young, and his papers are not referred to, because he didn’t discover anything about the rats. In fact, he discovered all the things you have to do to discover something about rats. But not paying attention to experiments like that is a characteristic of Cargo Cult Science.

Ensign Steve · #**458** 11-01-2021, 05:48 PM

What are robot dogs actually being used for? -...

Not gonna lie, I'm in love with robopuppy.

Limoncello · #**459** 11-01-2021, 05:52 PM

Those things creep me the fuck out.

Ensign Steve · #**460** 11-01-2021, 05:53 PM

We also keep chihuahuas, so I guess creepy-ass dogs are my jam. :chihuahua:

Kamilah Hauptmann · #**461** 11-08-2021, 10:42 PM

Saturday Morning Breakfast Cereal - VR

Kamilah Hauptmann · #**462** 11-10-2021, 06:55 PM

Ai+joke_1ceec8_9090610.jpg

Ensign Steve · #**463** 11-10-2021, 06:59 PM

That's cute an all, but face pareidolia was kind of a huge development in the evolution of human perception. JUST SAYIN

Kamilah Hauptmann · #**464** 11-16-2021, 04:53 AM

This is the best metaphor for machine learning I have ever seen. https://t.co/I11asLyrwp
— Alistair Croll (@acroll) November 15, 2021

Ensign Steve · #**465** 11-17-2021, 06:28 PM

If I wasn't already Team Skynet, I definitely am now! :cheer:

Researchers develop 'Minority Report' like tech for drugs - Vancouver Is Awesome

Quote:

Researchers at Vancouver's University of British Columbia (UBC) and other universities have trained a computer to predict the molecular structure of new drugs.

"There is an entire world of chemical ‘dark matter’ just beyond our fingertips right now. I think there is a huge opportunity for the right AI tools to shine a light on this unknown chemical world,” says Dr. Michael Skinnider in a press release. He completed the research as a doctoral student at UBC.

Designer drugs are newly designed drugs that skirt the law; they typically are very similar to previously developed drugs and have similar effects, but are different enough to not technically fall under the laws regulating the drug they were made to mimic, like new versions of bath salts and synthetic opioids.

Using data collected from around the world on illicit drugs, researchers trained the computer to come up with new drugs that hadn't been created yet, but that would fit the parameters. It came up with 8.9 million different chemical designs.

Then they compared 196 newly created designer drugs, that didn't exist when the computer was initially programmed, with those it had come up with.

The computer, a deep neural network, had come up with more than 175 of the drugs.