I’m an AI skeptic. While it can have its uses, I’m very concerned about the external costs being paid by our communities because of the rush to embrace AI. Power and water demands to run the data centers that AI systems require are increasing costs for everyone, and often those costs are falling disproportionately on those communities who are already at-risk and can least afford it. Some communities have seen data centers create noise pollution issues, while research has also shown that they’ve created heat islands. The Legislature considered a bill during the last session that would have required facilities that use high levels of power and water to publish reports that include information on their water and energy use, and for new facilities to use electricity from new renewable resources or nonemitting electric generation for 80 percent of their needs by 2030 and renewable or nonemitting electricity for 100 percent of their needs by 2045. I think we need to revisit the topic to ensure that those who are creating the increased costs are also responsible for paying them.

The training of systems with little regard for copyright protections is also very troublesome. I continue to see marketing that suggests that AI is a thinking tool, rather than a next generation predictive text generator. AI does not have a conscience. The value of any AI system depends on the data used to train it — garbage in, garbage out.

That said, AI is a part of our world and we have to approach it with our eyes wide open. Like many issues that rely on technology, AI is not immune to the digital divide. All our students need to have equitable access to the tools needed to build AI literacy. They also need careful instruction on AI literacy to effectively use those tools, similar to the work we’ve done on media literacy. We can’t separate those two. We need to be mindful that our AI tools are used to supplement our educational services, and not supplant them. We need to be very aware of the biases that can be present in the training of AI systems. There needs to be a systematic review of any AI program for potential bias before it’s allowed in the classroom. Biased input means biased output, even for careful users. If we keep those risks in mind, there can be value for our education system.

Outside the classroom, I’m concerned about the role AI might play in hiring and firing decisions. It’s a tool, not a manager, and should be treated as a tool and not a final authority. I’ve seen it said that on topics people aren’t familiar with, AI can seem 100% reliable, while on topics they are familiar with it’s only about 20% accurate. We’ve all seen stories of what’s been dubbed to be “AI hallucinations,” citing articles that don’t exist, court cases that don’t exist, or the infamous four and six fingered hands that were all too common in AI generated imagery. AIs will get more refined, but that doesn’t change concerns about their value and their role.