By Cory Gaines | Guest Commentary, Colorado Accountability Project
In last year’s legislative session there were 705 bills introduced, with 525 passed. I’ve not seen anything official, but the rumors I have heard have it that there are currently 200 bills in the queue for the 75th Legislature this year, with a prediction that the total number of bills could hit 600 to 700 by May.
What spurred my looking at bill numbers this year was an email forwarded from a friend. It was a marketing email by Colorado Capitol Watch (CCW), a group that bills itself as the “the premier” bill and legislator tracking site for Colorado. The email offered a novel service: using AI to make summaries of legislation.
Members of CCW are entitled to use their AI tool to help prepare abstracts in plain English for the bills they’re wanting to follow. This is quite a novel idea, and likely useful. Stacking the confusion of untangling legalese on top of bill numbers like we’ve seen the last two years (and likely more to come), would indeed make an assist welcome.
I don’t have access to CCW’s particular bill reader, but the idea intrigued me enough to look at a couple of bills myself. I used Google Gemini rather than have to register an account with the other AI sites. I input a couple of bills and asked Gemini to “turn this into plain English”.
The first bill was SB 25-144, “Change Paid Family Medical Leave Insurance Prog,” which I figured would be juicy with all kinds of complicated language and fiscal changes.
Gemini returned the following list of bullet points:
This Colorado Senate Bill (SB25-144) proposes changes to the state’s “Paid Family and Medical Leave Insurance Act.” Here’s a breakdown:
- Extended leave for NICU parents: The bill would allow parents of children receiving inpatient care in a neonatal intensive care unit (NICU) to take up to an additional 12 weeks of paid family and medical leave. This is in addition to the standard 12 weeks.
- Premium adjustments: The bill changes how the premiums for the paid leave program are calculated. Currently, the premium is 0.9% of an employee’s wages. This would remain in effect through 2025. In 2026, the premium would decrease slightly to 0.88%.
- Future premium calculation: Starting in 2027, the director of the division of family and medical leave insurance would set the premium rate annually by Nov. 1 of the preceding year. The director must set the rate so that:
- The fund has at least six months’ worth of projected expenses at the end of the year.
- The premium rate is stable (doesn’t fluctuate wildly).
- The premium never exceeds 1.2% of an employee’s wages.
- Effective date and applicability: The bill would take effect the day after 90 days following the end of the legislative session, unless a referendum petition is filed. If a petition is filed, the bill (or the challenged part) won’t take effect until it’s approved by voters in the November 2026 election. The changes to paid leave benefits apply to leave claimed on or after the bill’s effective date.
Not too shabby when you compare it to the human-generated summary from the bill’s fiscal note. Think of the brain damage you would avoid by feeding the bill into AI and getting something like this back.
I next turned to SB 25-001, a confusing bill which seems to not actually be geared toward solving any current problems. In florid and confusing writing, replete with definitions so dense they would sink in a pool of mercury, it outlines a series of “fixes” to voting problems our state is not having. If ever there were a bill ripe for AI dissection, this would be it.
I laughed a little when Gemini returned the following: “I can’t help with that right now. I’m trained to be as accurate as possible, but I can make mistakes sometimes. While I work on perfecting how I can discuss elections and politics, you can try Google Search.” I hear ya Gemini, I hear ya.
My results were pretty mixed in the two relatively simple trials I ran. I can’t help but think that CCW’s purpose-built utility, working alongside a human, would probably give much better summaries. But better AI is not the point we ought to be focusing on here. Using AI in and of itself isn’t either. Even assuming a perfect computer utility would be but to treat symptoms of a problem, rather than attacking the underlying cause.
There’s little that can be done about the legalese. One could make reasonable arguments that it has to be there. Language that is too loose would not be a good thing, especially when you consider the stakes of a misunderstanding or misinterpretation. Think about the other half of the problem, however. What on Earth do we need 600-odd new pieces of legislation for every single year? Do you find yourself, mid-June, thinking that what this state needs is more laws to run even better?
I know I don’t, yet the pattern seems to be set now, and all the computer tools in the world cannot get us past the fact that we’re human and can only attend to so many things at once (yes, this also includes our legislators who are supposed to be vetting these things). I’m glad we have AI to help us and I’m sure it will have a place in legislation, lobbying and citizen advocacy, but I hope we realize that the real problem to solve is not one a computer can handle.
Editor’s note: Opinions expressed in commentary pieces are those of the author and do not necessarily reflect the opinions of the management of the Rocky Mountain Voice, but even so we support the constitutional right of the author to express those opinions.