When AI is the tool that enables breakthroughs in cancer research and treatment which result in the eradication of cancer- will they still be a waste of resource?
Extrapolate that along any other deep-research-needed topic and ask yourself- "is this going to be improved by a tool which can have awareness of vast sums of information that can be queried by researchers using simple english language queries?"
There isn't a field that won't be impacted. Many already have been. Radiology is using AI to process X-rays and other diagnostics with a
detection rate much better than that of humans. Meterology is using it to
forecast the weather with greater accuracy. It's being used
EVERYWHERE.
I get the concerns about datacenters and the environment, the impact on power markets, water availabilty , etc. Those are all manageable impacts- they can be addressed by regulating datacenters rather than letting them operate as they do today- buying up huge swaths of power at discounted rates for years, joining municipal water systems and paying discounted rates to the system for the water they consume, all in the name of "incentivizing business growth" in cities (which never manifests, as "business" used to mean jobs, and now it doesn't.) Instead, tax those things- tax the heat they emit, tax the water they consume, tax the power they use- and over time, ratchet up requirements until they aren't a negative impact at all.
But to look at AI and say "we shouldn't even try this, because the planet" is like looking at the first man to create fire and saying "but should we burn the whole world?" and deciding that nah, fire's a bad idea.