Re: [sl4] Friendly AI and Enterprise Resource Management

From: Samantha Atkins (sjatkins@gmail.com)
Date: Wed Oct 06 2010 - 03:59:57 MDT


On Oct 5, 2010, at 1:11 PM, Mindaugas Indriunas wrote:

> It might be that one of the best ways to bring about the friendly AI is by trying to be very rational about one's own actions, defining one's own goal of life, and doing it in such a way that the resulting goal would be the objective good; and consequently designing a resource management system to automate and optimize one's own resource management decision making to achieve personal prosperity, which is identical to the prosperity of everyone.
>

Please define "good" and especially "objectively good";

We don't seemed to be wired to be particularly rational about all of our actions. So to what extent can this be taken and by what percentage of people?

What is your argument that your prosperity is identical to the prosperity of everyone? If you mean you being well off means automagically that everyone else is then I don't think so. Unless oneself becoming rich translates to optimal things be done by you for the benefit of everyone which are greater than any other uses that money could have been put to. If you mean you are not *really* prosperous unless everyone is, well I can almost see that if and only if everyone prospering in fact led to more prosperity (things I value) than could otherwise (and perhaps more readily) have been done. There is great heart value in everyone being prosperous but it is not necessarily to most rational way to personal prosperity. I have, over time, thought of many arguments for something of the kind so I would be very interested in what your reasons for thinking this turn out to be.

> Resource management is what almost every company is doing, and larger companies use sophisticated resource planning software for that. It is entirely possible to apply machine learning algorithms to automate increasingly many company decisions by analysis of data being collected in such systems, and it seems likely that that some hi-tech companies are already doing that, effectively creating AI-augmented corporations.

From the inside, you would be surprised how much very unsophisticated means are used and just how much waste of resources is endemic.

>
> However, corporations have been observed to conduct unethically, and not benevolently, and there is no guarantee of responsibility, sentience and friendliness of corporations in general. It is not entirely clear what particular goals is a particular corporation's automated business decision making systems are driven by.

What is "benevolent". It is benevolent, as a business, to create more actual value than was consumed in the creating of it. That is one of the most "objectively good" thinks I can think of. Is that part of what you grant is benevolent? The goal is receiving more tokens of value than you spend in producing what you received them for. Of course "tokens of value" are not directly value - especially if the token has become debased. A big problem is that we can't much believe in the value behind the tokens or that they are rationally allocated on the basis of actual value, even actual perceived value.

>
> Recently, I had an opportunity to see how resource planning software can be useful for a company's goal tracking, and to start working on my own system to help me manage my own resources to succeed in life.
>
> It is a very interesting field of research, because it is possible to adapt the newest machine learning algorithms to automatize one's own decision making process.

Cool. Yes it is interesting, and an area I don't have much time to explore in its fullness unfortunately. Is there a project brewing with some discussion and code base I might get involved in on some basis?

>
> However, instead of secretly doing one's own personal system to achieve personal goals, I think - for the sake of creating a friendly AI - it could be much better to develop such a system in open-source fashion, with a goal that's acceptable and desirable by all.
>
> I'd like to know, is anyone on this list is doing or thinking something along these lines?

Not at the moment but I have had some ideas and worked on a commercial product in the Knowledge Management area some years ago. I have some ideas on such personal information and decision software in its functionality and rough plans for how some of it could be done. There is actually a whole slew of projects begging to be done in this area.

- s

>
> Mindaugas Indriƫnas
> http://universians.org



This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT