Illustration by Laurène Boglio

Listen to this story:

Anyone who hasn’t been hiding under a rock has probably heard about the new tool ChatGPT. It’s a chatbot created by the San Francisco, California-based company Open AI that uses machine learning to generate human-like responses to typed questions or statements — sort of like an internet search engine that you can message with. Whether you think this new artificial intelligence (AI) is poised to ruin academic publishing, improve mental health care or threaten democracy, the fact is that people are using it.

The new technology isn’t drastically changing scientists’ lives yet: About 80 percent of respondents to a February Nature survey reported that they have tried ChatGPT or a similar tool, but most are using it just for fun — not for research.

But ChatGPT and other large language models like it hold enormous potential to make research easier, says Prachee Avasthi, co-founder and chief science officer of Arcadia Science, a company based in Berkeley, California, focused on innovating scientific research and development.

Avasthi first saw potential in using ChatGPT to craft email templates. She prompted it to generate an email to “politely” decline an invitation because of insufficient bandwidth. “It generates a perfectly reasonable template that takes two seconds to edit — an enormous amount of time saved,” she says. “We are underusing ChatGPT,” she wrote when she tweeted out the result.

We are underusing ChatGPT pic.twitter.com/PkiPaHLWn2

— Prachee Avasthi (@PracheeAC) February 4, 2023

Avasthi, who admits to being an early adopter of new technology in general, thinks of ChatGPT almost like a personal assistant. It’s great at generating copy, she says, or providing phrasing for internal documents. For example, she asked it to create a set of instructions for cleaning up common lab spaces, and it came up with a useful list that a researcher could tailor to their lab.

She says she was also impressed by the AI’s ability to troubleshoot an experiment involving PCR, or a polymerase chain reaction — a technique for copying DNA. When she asked it to explain why, in an experiment, PCR would generate too many bands on an agarose gel, it correctly suggested experimental conditions that could lead to that result, including contamination, primer design and enzyme selection.

Among other things ChatGPT did well, Avasthi says, was make a list of potential interview questions for a new lab technician, come up with ways to get discounts on lab equipment, list summer internship positions for Ph.D. students in the San Francisco Bay Area, and name granting agencies that fund cell biology research at universities in the United States.

Margaret Crane, a Ph.D. candidate in clinical psychology at Temple University in Philadelphia, Pennsylvania, has found another way to integrate ChatGPT into her life: using it to debug her code. “It’s been sooo much easier than trying to figure out with google,” she tweeted.

Crane has also found it to be helpful for brainstorming research ideas, she says. But of course, she adds, anything it spits out should be taken with a grain of salt. “It’s the most useful when you know a little bit about an area — or at least enough to know when it’s completely off base,” she says.

Avasthi agrees. Although there are potential pitfalls with the new technology, she says she sees it as a tool in the same vein as a calculator. If researchers can figure out how to use it for low-lift tasks, “we can put our energy and brainpower to other things.”

(For what it’s worth, when I asked ChatGPT how early-career researchers can use it to save time and resources, the bot had some decent but vague ideas, including literature reviews and paper editing — nothing as interesting as what real live researchers had to say! If you have any of your own tips, let me know at [email protected].)

Jobs, trainings and funds:

• King’s College London in the United Kingdom is accepting applications for two fully funded Ph.D. positions in the psychology department, including the option to work on a project to understand diverse life outcomes in autism.
• Early-career researchers who wish to conduct their work in the U.K. but are based outside of it can apply for a Newton International Fellowship, which covers funding for three years.
• Rebecca Shaffer, associate professor of pediatrics at Cincinnati Children’s Hospital in Ohio, is hiring a postdoctoral researcher in autism research and clinical care.

Recommended resources:

• The U.S. National Institutes of Health is requesting input on “challenges affecting retention in the postdoctoral trainee community,” according to a February tweet. Responses can be submitted through 14 April.
• Speaking of challenges, institutions need to better compensate trainees, writes H. Holden Thorp, editor-in-chief of the Science family of journals, in an op-ed. To afford higher salaries, he writes, academia needs to do “less science more humanely.”
• Permanent employment is another way to improve researchers’ working conditions, according to an article published this month in Nature Human Behaviour.
• Successful grants have a few things in common, including being clear and persuasive. Chaolin Zhang, associate professor of systems biology at Columbia University, who reviews National Institutes of Health grants, tweeted some advice for how to make an application stand out.

Tell what you would like to do in one or two sentences first, and then explain why this important, and how you are going to do it (pretty much like a news report). 1/n

— Chaolin Zhang (@chaolinzhang) February 12, 2023

• Retracting a paper was embarrassing, but it was also an opportunity to learn and grow, writes Jaivime Evaristo, assistant professor of geosciences at Utrecht University in the Netherlands, in an article for Science.
• It’s not always clear how to end a research project. Focusing instead on “deprioritizing” the work, redefining it or figuring out how to let go while transferring the knowledge to others may help with that tricky transition, according to a Nature article.
• Research faculty are inundated with requests for things that take time and might or might not benefit a person’s career. Russell Poldrack, professor of psychology at Stanford University in California, asked his Twitter followers which opportunities they’re glad they turned down early in their career — and those they wish they had. “I’m happy I said no to a high-demand associate editor position at a major journal,” he wrote.

I wish I’d said no to projects that were interesting but too far outside my core focus. Building the lab on established expertise would have helped me get grants faster rather than envisioning the lab I wanted in the future.

— Dr. Tracy Centanni (@tmcentanni) February 18, 2023

• And do you suffer from New PI brain?

New PI Brain pic.twitter.com/uc7ePZVTy4

— Sydney Trask (@Sydney_Trask) February 9, 2023

Helen Willsey, assistant professor of psychiatry and behavioral sciences at the University of California, San Francisco, tweeted that she’s currently intent on “reversing this ratio.”

Do you have advice for spending more time on science and less on worry? Share it at [email protected].

Cite this article: https://doi.org/10.53053/TYCH2095