Over the course of the next ten to twenty years various projects will enhance their systems to the point that they begin to exhibit intelligent behavior. What will the ethical framework and basic values be for these various systems?
- Search engines will be driven to identify spam and SEO chaff and to foresee what information you will want next, even before you realise you want it.
- Trading systems will be driven to maximise profit. Will they consider the collateral costs? Will their actions maximise long term value?
- Military systems, will be driven to identify threats and strategic opportunities. What is the basic function they will be seeking to maximise?
Next time some team of students develop an AI perhaps they could point it at Wikipedia to try it out. Instruct it to read the site content and derive the ethical basis and basic drives and then analise various resources and come up with useful contributions? Or maybe look at Wikiversity and intelligently prepare learning materials based on Wikipedia content. How would such an AI behave?
I propose that we should consciously consider what we do and say in our Talk: and Wikipedia: pages and act as if the AI that will one day run our world will take it's standard of behavior from the standards we are building here.
We should encourage AI researchers to use our data and advise us of ways in which our data can be better presented (but only if those changes also improve how information is presented to humans).
We have bots doing useful tasks now. We should continue to encourage users to develop better bots, and continue to develop the framework within which they work as necessary.
Even if no AI ever gets to run the world I think that acting as if it will, will lead to better behavior. Don't bite that Newbie - he may be your new wikimaster!
Encouraging AI researchers to work with us will help to frame and shape how AI develops and thus will help shape our world for a long time ahead. We don't want all AIs thinking like Goldman Sachs or the DOD.
- Are our policies and values clear?
- Does our conduct live up to our rhetoric?
- If the whole world ran the wikipedia does would the world be a better place?
- I there a limit to how much intelligence our bots should have?
Some thought and care.
Do you have a thought about this proposal? A suggestion? Discuss this proposal by going to Proposal talk:Wikimedia projects rules of behaviour as the basis for a system of ethics for an Artificial Inteligence.
Want to work on this proposal?
- .. Sign your name here!