While We Debate the Future of AI, Are We Missing What’s Happening Right Now?

by Claire L. Brady, EdD

I read a fascinating WIRED interview with AI researcher Timnit Gebru this week, and I haven’t stopped thinking about one part of her argument.

Gebru is deeply skeptical of the increasingly loud conversation about AI as an existential threat to humanity. She argues that all the talk of superintelligence, rogue AI, and machines potentially wiping us out can distract us from harms that are already here: environmental costs, labor exploitation, data and copyright questions, lack of transparency, and our tendency to over-trust systems that produce remarkably convincing answers. Pasted text

You don't have to agree with Gebru's assessment of existential risk to take that challenge seriously.

The Future Is More Interesting Than the Present

Because I think higher education has its own version of this problem.

We spend a lot of time trying to anticipate what AI will do. Will it replace jobs? Will agents transform how universities operate? Will students need entirely different skills? What happens when AI becomes dramatically more capable?

Those are important questions. But sometimes the future is so much more interesting than the present.

Meanwhile, AI Is Already Here

Students and employees are entering institutional information into AI tools today. Faculty are trying to determine whether student work is authentic. Universities are signing contracts with vendors whose models and data practices may be difficult to understand. AI-generated information can sound authoritative while still being wrong. Data centers are creating real questions about electricity, water, and infrastructure. And institutions are trying to make decisions about technologies whose capabilities are often described by the same companies selling them.

Impressive Is Not the Same as Trustworthy

Gebru is particularly concerned about what she calls automation bias: our tendency to over-trust automated systems. Fluent, confident output can make an answer feel more reliable than it actually is, especially when there are few visible cues that something has gone wrong. That one should be circled in red for higher ed. The more capable AI becomes, the easier it is to confuse impressive with trustworthy. Our responsibility isn't simply to teach people how to use these tools. We also need to help them develop the judgment to question them.

So what does that look like for higher ed leaders?

Ask what is happening now, not only what might happen next. Your AI governance conversations should include current use, current risks, and current impacts alongside emerging capabilities. What tools are already being used? What data are going into them? Where are people relying on AI-generated information in consequential work?

Demand more than impressive demos. Ask vendors about training and institutional data, retention, privacy, documentation, evaluation, human oversight, and known limitations. Gebru argues strongly for greater transparency and documentation around AI systems. Pasted text Higher ed should be asking for exactly that before putting these products into consequential workflows.

Teach skepticism alongside skills. AI literacy cannot stop at prompting. Students and employees need practice verifying outputs, recognizing uncertainty, understanding when human expertise is required, and knowing when not to use AI.

Broaden who is at the table. AI isn't only an IT or an academic issue. The questions Gebru raises touch sustainability, procurement, academics, student affairs, HR, legal counsel, research, equity, accessibility, communications, and institutional risk. Our governance structures should reflect that reality.

There is a lot we genuinely don't know about where AI is headed, and I don't think higher education should stop thinking seriously about the future.

But Gebru's challenge is an important one: don't let speculation about tomorrow keep us from taking responsibility for today. We don't have to settle every debate about the future of artificial intelligence before we make better decisions about the AI already on our campuses.

Image created using ChatGPT