Don’t bet on AI too soon

By Raena Hunter Doty Staff Writer The future of artificial intelligence is uncertain. I say that, as much as I can, without an agenda or bias. My opinion on AI is complicated, multi-faceted, and at times contradictory. I will be the first to admit this. People on the internet will sound certain when they talk about AI - particularly AI chatbots based on large language models, such as ChatGPT or Google’s Gemini. They will tell you that AI is single-handedly causing the destruction of the world or that it is single-handedly saving the world. They will tell you that it is a tool that’s powerful when used correctly or destructive when used at all. They will tell you that it is making people’s jobs easier or that it is destroying people’s livelihoods. But these claims are made over-confidently. People who make them are often people who have spent little time contemplating the other side of the story, and they cherry-pick often anecdotal evidence in order to affirm what they already believe. The truth? AI is constantly evolving. New models are published frequently, constantly changing what is possible with AI - for better or worse. Our collective understanding of AI - what it is, what it can and cannot do, and how to use it most effectively - is always progressing. Products utilizing AI are becoming more specialized. At the same time, many of the questions people asked after ChatGPT released in November 2022 are still unanswered. Can we justify the immense environmental cost of AI tools? Should it be allowed for AI tools to be trained on the creative products of people who didn’t consent to companies using their works in such a way? Can we trust machines that routinely generate false information because they aren’t actually intelligent, they just seem to be so? Can we possibly trust machines that perpetuate the biased, often discriminatory nature of everything in their datasets? I know I sound like a broken record here. You’ve probably heard all of these arguments a million times, and if you have a strong opinion one way or the other, you’ve found a way to justify dismissing the other side. Don’t take my words as judgmental. I find it hard not to pick a side and stick with it until the end, too. For a long time, I was staunchly anti-AI. Even now, the thought of using it myself makes me feel gross, and I avoid doing so as much as possible. But I’ve been inundated enough with suggestions of how to use AI - use cases, best practices, strategies for most effective results. Now, when I face a problem, I often find my mind drifting toward the thought - “I could just type this into ChatGPT. It would be easy.” I understand the temptation, and I’ve even read enough testimonials to know the results can be quite effective. I trust myself enough to believe that I would use AI well, if I used it at all. So despite my personal hesitance to use AI, I don’t judge as much as people often expect. Regardless of your personal AI use - whether you use it every day or have never even touched it - I call for you to accept one central premise. The future of artificial intelligence is uncertain. Everyone will have a different baseline amount of AI usage that they consider acceptable and helpful for themselves, but as a society, we need to leave room for this uncertainty. So whoever you are - whether you’re a student writing an essay, a professor writing a syllabus, or an administrator writing an ethics policy - consider one question: “What if I’m wrong? If I’m wrong, how will my actions now affect myself and others in three years, five years, 20 years?” This can be interpreted many ways. It means you shouldn’t become dependent on AI. It means you shouldn’t disregard it so heavily that you’ve shot yourself in the foot when you make your next career move. It means you shouldn’t codify anything that will be difficult to undo. Right now, the entire world is in a transitory phase because of AI - and that’s OK. We can lean into the transition. Lean into the fact that the future of AI is uncertain.