Friday, April 12, 2024

Ethical AI Use

I train people to use technology as their primary job. Everyone thinks there are some kind of "copyright issues" associated with AI use, but that just doesn't appear to be the case. 

Reading a book or news website, or using it to train yourself to understand English, so that you become marketable, is perfectly acceptable use. In the same way, using a book or news website to train an AI is really no different than using those same books or websites to train a non-native language speaker to understand and speak the language so s/he can get a better job. No copyright infringement there.

The US Copyright office does not recognize machine-output as something that can be copyrighted. So, using AI output as your own is perfectly legitimate. AI is a tool, like a printer. If I give a printer a text document, that document gets processed through the application software, the OS, the printer drivers and finally the printer, before the user gets a printed page, yet we don't claim the user is somehow being unethical when they pass off that printed page as their own work.

Yet that printed page is just as much an individual's own work as the output that comes from an AI after the user provides a prompt. Tens of thousands of people, millions of man-hours, came together so that the application, the OS, the drivers and the physical printer would do what the user wanted. Same thing is true of AI. Neither will do anything until the user prompts them to do it. 

There are really no ethical or copyright issues involved in the use of AI. But, people don't want to hear that, so they shut their ears to it. 

I discussed AI with a student. She could see that it would overtake various aspects of the technological job we train our students for. She observed that an AI could program a router or switch, or entire classes of devices, much faster than she could. 

The advantage she felt she had was in orchestrating the overall purpose. While AI is good at grunt work like generating essays or programming interfaces, it doesn't do it for a reason, but only in response to user prompts. She didn't see that changing anytime soon.

She is certainly correct about the AI getting the step-by-step recipes done faster and more accurately. It's not clear to me whether she is correct about the AI not seeing the overarching purpose of a specific AI-curated event or situation.

This borders on the realm of the philosophical. It depends on whether or not purpose is ultimately a natural (i.e., mathematical) result or if it depends on some supernatural level of understanding. In short, this is really a question revolving around whether or not you believe in God.

If mankind is purely natural, and if math is truly the language of nature, then AI will be able to attain a truly purpose-driven perspective on what it does. In this perspective, AI and man are equally natural, and therefore can be equally powerful.

However, if mankind possesses a supernatural element, a strand in his/her makeup that transcends the natural/mathematical realm, then AI will not be able to capture the essence of "purpose." If man is truly a possessor of the "spark" of the divine, or somehow made "in the image and likeness of God", then AI - a natural, man-made artifice - cannot leap across that infinite chasm.

It will be interesting to watch how this pans out. 

 

No comments:

Post a Comment