You’ve likely come across a recent news story about ChatGPT, a new artificial intelligence tool developed by OpenAI that produces human-like text in response to questions and other requests.
There are reports of college professors around the country scrambling to make sure their students aren’t using ChatGPT to complete their assignments. A little closer to home for me, a nuclear energy lawyer, I came across a SCOTUS blog illustrating that ChatGPT isn’t very accurate at answering questions about the Supreme Court.
But I wondered: how would ChatGPT measure up on nuclear energy, as well as legal and regulatory concepts important to other practitioners in my field?
The answer is: not very well. At least not yet.
ChatGPT is fun to use and can generate some very useful, basic information. For example, it gave thorough and understandable answers to questions like what is nuclear energy?, what is the U.S. Nuclear Regulatory Commission?, and what is the Atomic Energy Act?.
And I can confirm it is correct about two things: nuclear energy is clean and reliable. As ChatGPT says,
| Is nuclear energy clean and reliable? |
|---|
| “Nuclear energy is often considered clean in terms of greenhouse gas emissions, as the nuclear reactions that take place in the reactor do not produce significant emissions of carbon dioxide or other greenhouse gases. This makes it an attractive option for countries that are looking to reduce their carbon footprint and mitigate the impacts of climate change. As for reliability, nuclear energy is known for being a highly reliable and consistent energy source. Nuclear power plants can operate continuously for long periods of time, providing a constant supply of electricity even during periods of high demand. This makes them an important part of many countries' energy mix and a valuable tool for ensuring energy security.” |
Other questions, however, were badly botched. I asked what the largest source of carbon-free energy in the U.S. is, and ChatGPT authoritatively said “in 2021, hydropower accounted for about 7% of total electricity generation in the United States and was the largest source of carbon-free electricity in the country.” As part of that response, it also noted that “nuclear power is also a significant source of carbon-free energy in the United States and accounted for about 20% of total electricity generation as of 2021.”
Now, a sharp third grader (or at least mine) can easily spot the mistake. And when provided with a leading question, ChatGPT corrected its mistake.
| Isn't 20% greater than 7%? |
|---|
| “My apologies, I made a mistake in my previous response. Yes, you are correct, 20% is indeed greater than 7%. Nuclear power is the largest source of carbon-free energy in the United States, followed by hydropower, and then wind power. Thank you for bringing this to my attention.” |
Given that the objectively correct answer was embedded in its initial response, I was left wondering if there is some sort of anti-nuclear energy bias within ChatGPT. It’s an interesting question that I’ll leave for those with experience in the intricacies of machine-learning algorithms. That’s not me, so I moved onto probe other topics of interest to nuclear lawyers.
Courts have recently been grappling with the issue of which federal officials must be removable by the President for any reason and which can have additional protections, allowing them to be removable only “for cause” (e.g., violation of law or abuse of authority). It’s an interesting question and involves some important constitutional considerations. Thinking I was pretty sure the answer was “no,” I nonetheless proceeded to ask if an NRC commissioner had ever been removed “for cause.”
| Has an NRC commissioner ever been removed for cause? |
|---|
| “Yes, a Commissioner of the U.S. Nuclear Regulatory Commission (NRC) has been removed for cause in the past. In 1989, President George H.W. Bush removed Commissioner Sylvia Lowrance from the NRC, citing her opposition to the Administration's policy on nuclear waste disposal.” |
I had never heard of a “Commissioner Lowrance,” but to be fair to ChatGPT, I was in middle school in 1989. Yet as someone who has been practicing in the field for nearly 20 years, call me skeptical. Even so, given the detail and confidence in ChatGPT’s description, I found myself checking the NRC’s list of former NRC commissioners before posing another leading question.
| Sylvia Lowrance was a commissioner on the U.S. Nuclear Regulatory Commission? |
|---|
| “My apologies, I made an error in my previous answer. Sylvia Lowrance was not a Commissioner on the U.S. Nuclear Regulatory Commission (NRC). I apologize for any confusion I may have caused.” |
So, that’s that. Despite being completely false, the initial response sounded so convincing. ChatGPT’s designers recognize these types of limitations and acknowledge that it “sometimes writes plausible-sounding but incorrect or nonsensical answers.” And they say they’re working on addressing these challenges. In the meantime, it’s a fun and innovative tool.
ChapGPT put it best:
| Can you prepare a blog article by a nuclear energy attorney who tried chatgpt? |
|---|
| “Overall, my experience with ChatGPT has been a positive one. While it's not a replacement for a human attorney, it's certainly a helpful tool that can save time and effort. If you're a nuclear energy attorney who is looking for a new and innovative way to work, I would definitely recommend giving ChatGPT a try.” |
Just remember to check your work, cite your sources, and have a questioning attitude about what it produces. Not everything you read from AI, or elsewhere about nuclear, is true. Especially if it says nuclear isn’t our largest source of carbon-free energy, because it is.
