More Thoughts on AI as a Tool for Good or Ill

Like getting a new bicycle in the 1890s (times a million), new opportunities for human advancement are suddenly available. A sense of freedom and open space without horse manure. A chance to grow in new and important ways.

Simultaneously, AI offers virtually limitless opportunities to consolidate wealth and power.

AI amplifies human capabilities and computer algorithms in ways that defy comprehension. And the number of different models, each built for a purpose, is accelerating. The biggest tech bucks are now on the line, and the hype is following close behind. OpenAI raised $300B in their latest funding round — more than the GDP of both Greece and New Zealand.

Sorry, but it’s still just a tool.

And like all tools, AI can cause harm, but on a global scales and millisecond time frames.

From influence peddling (extremely popular) to harassment, extortion and drone strikes, AI is optimizing, if not outright controlling, our physical, real world tools. Think of Boston Dynamics robots dancing and doing gymnastics. 

• How will this AI instance help your stakeholders (broadly defined)?
• What harms could it cause those same stakeholders, and how are they mitigated?
• What values get encoded into the learning model, and do they lead to healthy outcomes for stakeholders.  For years, ad agencies have individually targeted each of us by tracking our online activity. Can you imagine how AI systems will leverage that? 
• Does it embed deceptive or extractive values/priorities?
• Who stands to gain, and is the desired ROI reasonable?
• Why are we doing it?
• Is it revokable if it proves harmful?
• What if it’s highly damaging while making boatloads of money?

Remember that the Internet started as an idealistic dream of information democracy. Now you encounter paywalls and advertising that obliterates content, and content itself that’s so bad that it should be used as toilet paper (figuratively speaking).

Already vigilant about malware, for examples, users must now be careful of emails that sound like they’re from Mom. Can you tell?

Like all tools, each AI model is a reflection of its designers and the totality of its operation: computer hardware, data sources and vast amounts of electricity.

To prevent likely harm, and assuming good intentions, start with the basics: What’s the purpose of the AI, how is it trained, on what data, and what controls are in place to keep it from causing harm?

Power tools, kitchen appliances, and bikes all come with prominent safety warnings. In contrast, and in a way that covers legal liability but is useless to the end user, AI’s typical warning notice reads “AI is experimental.” Thanks a lot.

Artificial, like Splenda, and intelligent only as an extension of ourselves.

It’s a fabrication, like canned soup, wicker chairs, and toaster ovens. Useful, lifesaving even, but still an artifice. It’s not conscious the way my dog is conscious.

And “Intelligent” is simply the wrong word. Generative more accurately describes what the systems are doing, but now we have “Generative AI,” which still has the sticky word in it.

We don’t actually know what “intelligence” is, human or otherwise. Or it has meant so many different things over time that it has lost its meaning. No efforts to objectively measure intelligence have ever succeeded.

Also, and I haven’t seen this much, the word intelligence applied to anything artificial demeans human beings because, let’s face it, these systems didn’t invent themselves.  There’s this weird idea that the machines are “better” than us.

Going back to the new bicycle of the industrial revolution, AI is a tool that amplifies human capacities. Unlike a bike, however, once deployed, AI (and all IT systems) roll merrily along in whatever direction its designers set it in. Make good decisions up front.

Finally, tech is not conscsiously intelligent so it’s unlikely that an AI could become a “singularity” — a mega being that realizes its own consciousness and wipes us out ‘cuz we’re stupid.

In sum, AI can be used for good or ill. Like a table saw, it cuts a board or a finger with equal indifference. There is no person, being, or consciousness in there, just a bunch of wiggling electrons.

So, now that we have our shiny new bike, where do we want to go, and why?