Fifty years ago, Cambridge faced down a frightening new technology and chose to write the rules instead of running away. That decision built the biotech capital of the world. We can do the same thing with AI.
AI is moving faster than any technology we have ever seen. The amount of work these systems can do on their own is roughly doubling every few months, and its pace shows no sign of letting up.
A lot of what is coming will be good for us. Faster ways to find new medicines. Real expert help for families who can't afford a lawyer or a specialist. But a lot of it is also worrying. These tools have already been used to help plan cyberattacks, and our schools are struggling to keep up with them. And it's too hard to predict what all of this might mean for your own job.
We can write good rules while this technology is still in its infancy, and Massachusetts is exactly the place to do it.
In 1976, Harvard wanted to build a lab in Cambridge for an unnerving new technology called recombinant DNA, the foundation of modern genetic engineering. Neighbors packed City Hall with objections, while the mayor stood up and warned about Frankenstein.
Cambridge did not panic or look away. The city paused the research, put regular residents on a review board next to the scientists, and then passed the first law anywhere in the world that governed this kind of research. That ordinance is still on the books today.
The clear rules gave companies the certainty to build here, growing Kendall Square into the biotech capital of the world. What started in our City Hall spread outward, to Princeton, San Diego, Ann Arbor, and beyond, and someone even wrote a book about how it all happened. A lot of life-saving treatments today trace back to a city council that chose to seize the moment, writing the rules instead of running from the future.
We can do the same thing with AI.
California, New York, and Illinois have already passed the country's first AI safety laws. They require the largest AI companies to publish their safety plans, report serious incidents, and protect the employees who speak up when something looks dangerous, and Illinois now brings in outside auditors to make sure the work is actually being done. Massachusetts should, at the very least, hold ourselves to those same standards.
And these things tend to spread: New York borrowed from California, Illinois used a bit of both, and Congress is now using all three laws as the starting point for a national framework. Beacon Hill is debating a version right now, and on this one the public is out ahead of the politicians.
Right now the law only asks companies to describe their own safety plans. We should set actual requirements the most powerful systems have to meet before internal deployment, including locking the models down so they cannot be stolen or tampered with. We crash-test cars and run trials on new medicine. It's common sense to do this for AI, too.
If a company's AI wreaks havoc or causes a catastrophe, that company should be on the hook for the harm.
The most powerful models should be tested by qualified outside experts before the rest of us ever use them. On something this consequential, companies can't grade their own homework.
If these systems create the kind of wealth their builders keep promising, working families need to share in it. When companies profit from automating the work that people used to do, some of that gain should return to the public that made it possible.
I care a lot about getting the rules right because the range of outcomes here is enormous.
AI puts half the country out of work.
AI lets us do the same work in half the time, for the same pay.
On an economist's spreadsheet, those two futures look almost identical, because in both cases the same work gets done with far fewer hours. In the lives of actual people they could not be more different. One prospective future hollows out families and whole towns. The other doesn't sound half bad.
Technology will keep moving no matter what any of us do about it. Policy is the tool we use to ensure we'll end up better off as technology accelerates. That, to me, is the entire point of doing this job well.
There is not a single engineer in the Massachusetts State Senate, and AI is moving so fast that even the people working on it are struggling to keep up. I studied engineering at MIT, I have built AI products myself, and I have spent years alongside the AI researchers and advocates shaping this policy. I don't claim to have every answer, because nobody does yet. But I do think I bring a particular insight to these questions and a real willingness to do the hard work of getting the answers right.
Fifty years ago, our community looked at an unsettling new technology, wrote the rules that governed them, and wound up leading the very industry it was afraid of. That is the Massachusetts way, and it is the work I want to take to Beacon Hill.