Technology:  Will AI be the death of us?  Maybe.

While catching up on reading, I came across a brief but chilling article in Time titled: “New Lab Partner: Why AI Models Could Help Prevent – or Cause – the Next Pandemic.”

The article cites a recent study where researchers asked AI models to troubleshoot highly complex lab procedures.  The results?  Ph.D.-level virologists scored just 22.1% in their area of expertise.  OpenAI’s O3 model scored 43.8%.  Google’s Gemini 2.5 Pro scored 37.6%.

Let that sink in.  Anyone with internet access now potentially has a nonjudgmental, 24/7 “virology assistant” capable of guiding them through the steps of creating a bioweapon.

While AI companies tout “guardrails” to prevent misuse, what about the platforms that don’t care — or are controlled by rogue nations?

Yes, AI will almost certainly deliver incredible benefits — cures, vaccines, climate modeling. But it also introduces unprecedented threats.

I’ve long warned that AI will upend the labor market, possibly displacing 20–48% of today’s white-collar jobs.  We’ve already seen what happens when blue-collar workers are left behind by automation and globalization.  The societal and political aftershocks — rising extremism, nationalism, authoritarianism — are still with us.

Now add to that the risk of AI-enabled bioterrorism.

What to do?   1.)  At the very least, we need serious contingency planning — now — for how we’ll respond if an AI-created bioweapon “goes off.”   2.)  We need a plan to deal with another tsunami of under- and un-employed workers whose jobs will be nuked by AI.  3.)  We need serious, enforceable global AI governance.

The clock is ticking.  Our elected representatives, policy makers, and think tanks need to get on it.   And the media needs to raise the public’s awareness of the issues.

PS.  I am working my way through a certificate program in AI.  The course encourages us to use generative AI tools in our daily lives.  Ever the obedient student, I did that here.  I wrote this blog post and then ran my version through ChatGPT, asking it to make it better.  It did.