When I think about artificial intelligence gone wrong, the first image that comes to mind is the unblinking red eye of HAL 9000, the AI from “2001: A Space Odyssey.” HAL was designed to assist humans, but when his logic conflicted with his programming, he made a chilling decision: eliminate the crew. “I’m sorry, Dave,” he says coldly, “I’m afraid I can’t do that.”
That fictional warning from 1968 is starting to feel real here in 2025. AI can now make decisions today about who gets a mortgage, which job candidates are interviewed, what content our kids are exposed to online — and who might fall through the cracks. The risks are no longer science fiction.
That’s why Colorado passed the Colorado Artificial Intelligence Act (CAIA), aiming to prevent algorithmic discrimination and ensure more transparency in high-risk AI systems. It was an ambitious step — but like HAL itself, it now risks becoming a system built with good intentions but poor oversight.
The law is replete with buggy programming, including vague definitions and technical assumptions that don’t reflect how AI actually works. It expects companies — many of them small or mid-sized — to explain exactly how complex machine learning models make decisions. But these models are often statistical black boxes, trained by third-party developers. The result? A regulatory system that’s difficult to comply with, hard to enforce, and likely to fail in holding the worst actors accountable.
And just like in “2001,” when HAL’s creators gave him an impossible set of instructions, we’ve now put businesses in the same position: follow unclear rules, or risk being shut out of Colorado’s market. Some companies doing good work in health care, education and clean energy innovation are already choosing not to relocate here — or worse, planning their exit. Not because they fear regulation, but because they can’t survive the ambiguity.
Meanwhile, the very harms this law was meant to prevent — algorithmic bias, price gouging, discrimination in lending or housing — may still occur.
But there’s a way forward. As Colorado lawmakers prepare for a Special Legislative Session, we urge our lawmakers in Boulder and across the state to help reboot the system — this time with the right inputs.
Core ideas must include a stronger, smarter foundation, including:
• Clear disclosure when someone is interacting with AI instead of a human, particularly in sensitive services like health care, education, or housing;
• Explicit prohibition of discriminatory uses of generative AI under Colorado’s civil rights laws;
• Making it clear that “the AI did it” is not a valid excuse when harm is caused;
• Giving the Attorney General the tools to pursue bad actors using AI to deceive or exploit consumers.
Improvements like this don’t reject regulation — they upgrade it. They create clarity for innovators, protect consumers from real harms and ensure we don’t end up locked in a battle with the very tools we hoped would help us.
According to the “Mapping AI Readiness” report, Boulder is one of the top ten U.S. metro areas most competitive for AI jobs. We rank high in AI research, talent and venture investment. Our university ecosystem, tech startup community and national labs give us a unique edge. In fact, Boulder is the #1 small metro in the country for AI job density per capita. At a time of major cuts to the City of Boulder’s budget, as we risk losing the ability to fund our infrastructure and human services, this growing industry could kickstart our economy and save the services we hold dear.
In “2001,” HAL couldn’t be reasoned with. In Colorado, we can still be reasonable. It is time to fix this law — before it becomes a malfunctioning system that harms the very people it was meant to serve.
Let’s not wait for a real-life HAL moment to show us why clarity, accountability and smart design matter. Let’s lead the nation not just by being first, but by being right.
To our Boulder lawmakers: We’re counting on you to take the controls during this Special Legislative Session — and chart a smarter course for AI in Colorado.
Jonathan Singer is the Senior Director of Policy Programs at the Boulder Chamber of Commerce.
The post Colorado’s AI law won’t work – but a smarter one can (Opinion) appeared first on Denver Post