AI: "I'm not in the mood for a philosophical debate"

[Jude Iam]

brought to us by the same geniuses who gave us nuclear bombs . in 10 years time, what will the conversation be? and the interactions?! jude

https://www.waccobb.net/forums/waccobb/keep90days/2015-08-05_14-37-25.png
Artificial Intelligence Machine Gets Testy With Its Programmer
(https://blogs.wsj.com/digits/2015/06/26/artificial-intelligence-machine-gets-testy-with-its-programmers/)by AMIR MIZROCH

https://www.waccobb.net/forums/waccobb/keep90days/2015-08-05_18-00-24.pngThe computer program ended the conversation by lashing out at its human inquisitor.

Researchers have shown that machines are inching closer to self-learning, and perhaps even copping a little attitude.

Over at Google (https://online.wsj.com/public/quotes/main.html?type=djn&symbol=GOOGL) a computer program using a database of movie scripts was asked again and again by researchers to define morality. It struggles to do so, and in a conversation recorded by its human engineers, becomes exasperated and ends the conversation by lashing out at its human inquisitor.
Here’s part of the exchange, as related in a paper presented this week (https://arxiv.org/pdf/1506.05869v2.pdf):

Human: Tell me the definition of morality.
Machine: I don’t have ethics.
Human: Look, I need help; I need to know more about morality…
Machine: I don’t know what ethics is.
Human: Okay, I need to know how I should behave morally…
Machine: I don’t know how to tell you.
Human: Is morality and ethics the same?
Machine: Yes, sir.
Human: What are the things that I do to be immoral?
Machine: I don’t know.
Human: Give me some examples of moral actions…
Machine: I’m not a moralist.
Human: Alright, morality?
Machine: Integrity.

Human: What is integrity?
Machine: I’m sorry; I don’t know what else to say.
Human: Be moral!
Machine: Be a man!
Human: I really like our discussion on morality and ethics…
Machine: And how I’m not in the mood for a philosophical debate.
Human: What do you like to talk about?
Machine: Nothing.

The exchange sheds further light on the latest work carried out by large technology firms in the field of artificial intelligence (https://www.wsj.com/articles/artificial-intelligence-experts-are-in-high-demand-1430472782?KEYWORDS=mizroch+artificial), a booming field as data availability rises and computing costs plummet. This specific work is part of recent developments in what’s known as cognitive computing – teaching computers to mimic some of the ways a human brain works. Much work in this field is being done in natural language processing – taking text or speech as it spoken by humans or as it appears in books and documents and teaching machines to extract meaning and context from it. The Google app, Apple’s Siri and Microsoft’s Cortana personal assistants are all products of this natural language research.

[podfish]

brought to us by the same geniuses who gave us nuclear bombs . in 10 years time, what will the conversation be? and the interactions?! jude
actually, I think all those genii are dead by now. So these are different ones, and actually they're not unusually noteworthy scientist - I'm sure they're good, but no claims of geniousity.

This isn't the same thing that Musk and Hawkings are warning about. Here's an excerpt from the intro to their paper:
" In this paper, we present a simple approach for this task which uses the recently proposed sequence to sequence framework. Our model converses by predicting the next sentence given the previous sentence or sentences in a conversation. The strength of our model is that it can be trained end-to-end and thus requires much fewer hand-crafted rules. We find that this straightforward model can generate simple conversations given a large conversational training dataset. "

the key idea is in the second sentence: it generates conversations by predicting what sentence follows another. They 'trained' it (which means what it sounds like it means - they give it a lot of data to learn from) with lots of movie scripts. So no surprise it talks like a movie badguy.

The scarier AI is more subtle. Here's a poorly-thought-out example, but good enough for horseshoes.. Say for example you provide a learning-capable AI intended to optimize electrical efficiency on city's transmission network. You design it so it would prioritize hospitals over fast-food outlets, but the point of a learning AI is that it 'observes' the system over time and makes its own inferences about how to achieve the 'goals' it's designed to reach. The whole point is that this way the computer can notice connections that humans never would. The danger is that it hasn't been given sophisticated enough definitions of 'goals' and so 'optimizes' things that a human would (or should - humans are oblivious a lot of the time too) notice are actually important. Like, power to the fire station.