Can A.I. Mean?

Listen to Episode No.4 of All We Mean, a Special Focus of this podcast. All We Mean is an ongoing discussion and debate about how we mean and why. The guests on today's episode are Bill Cope and Mary Kalantzis, professors at the University of Illinois. In this episode of the Focus, our topic is whether A.I. can mean.

The short answer is yes, A.I. can mean... whatever we make it mean.

For instance, ChatGPT does has access to text on certain kinds of subject matter, like, for example, the assembly of explosives or specifications on suicide. This kind of stuff is on the web, so ChatGPT has “read it” these subjects into its corpus. However, human programmers have applied filters telling the A.I. not to speak about these things. Nonetheless, you may be able to get to what it “thinks” about these things with some clever prompts, called “jailbreaks” in the hacker trade.

But does the A.I. really think, as we humans would associate with the act of thinking? Not really, because an A.I. like ChatGPT does not think about bombs or self-destruction. It just has words about these subjects which it doesn’t itself “understand.” And on top of that, its human-programmer masters have told it not to repeat them.

But whose purpose is meant to be served here, the A.I.'s or our own? In our discussion in this instalment of All We Mean, we argue, of course, for the A.I. serving the purposes of us humans. But there the questions immediately arises, which of us humans will be served? It may be that only the big stakeholders in the large Internet companies get served, and who knows what purposes they have. Perhaps they're quite content to see A.I. create the illusion of fact and consciousness, if for no other reason than to increase profits.

We, on the opposite side of that, say that the technology has tremendous potential for everyone, if used in everyone's interests. For example, people who want to learn can use A.I. technologies to improve their own performance, just as people who want to discover can use A.I. technologies to communicate their findings more effectively. These are the sorts of purpose we believe A.I. can be used to accomplish, by anyone, for everyone. But, we wonder, will purposes such as these also count when the technology rests firmly in the hands of the very few, because what if they don't really care what the rest of us want?

Read Bill's and Mary's multimodal grammar of A.I.

And read their work on using A.I. in education.