A growing sense of pessimism about artificial intelligence is spreading through the scientific community. While some AI experts remain optimistic, many researchers express significant concern over its negative impacts, from the proliferation of misinformation and AI-generated “slop” to its immense energy demands, exploitation of workers, and the consolidation of power by Big Tech. Surveys indicate that for every scientist excited about generative AI, nearly three are concerned.

This widespread negativity presents a unique danger. If the experts best equipped to guide AI’s development view it as a lost cause, they may disengage from the process entirely. This leaves the technology’s trajectory to be shaped without their critical input, making harmful outcomes more likely. Just as with climate action, a compelling, positive vision is needed to galvanize efforts toward beneficial goals.

Charting a Positive Course for AI

Fortunately, a path toward a more positive AI future already exists. The technology is being used to break down communication barriers for marginalized languages, accelerate major scientific discoveries like protein folding, and even help policymakers better engage with their constituents. These nascent examples prove that AI can be wielded for the public good.

To build on this momentum, scientists and engineers are encouraged to take four key actions. They must work to reform the AI industry to be more ethical and trustworthy, resist harmful applications by documenting and exposing them, responsibly use AI to solve societal problems, and advocate for the renovation of institutions to prepare them for AI’s disruptive impact.

Ultimately, technology is not inherently good or bad, but its direction depends on the choices made today. Scientists have a special responsibility to not only warn of the risks but also to champion and build an AI-infused world that is equitable, democratic, and beneficial for humanity.

Source: https://www.schneier.com/blog/archives/2025/11/scientists-need-a-positive-vision-for-ai.html