AI for scientists

We asked our community members which AI tools they use, what they are helpful with, and what they can suggest to other community members.

Compilation and editing by Olya Vvedenskaya

Work with literature

Our community scientists use AI tool Claude.ai as a chatbot. With the help of this tool, they prepare short letters, analyze articles, and write summaries. They add that they, however, always check the output.

Another tool mentioned in the community is Perplexity.ai which is a search engine with AI. It is claimed to be some sort of an improved Google, it does a sort of summary and adds references and sources.

Another researcher says that they use the tool Notion AI. ‘I tried some models that help to find relevant literature and some generative models’ - says one of the community members. ‘While Notion AI can’t help much with generative models, it can read PDFs and work with tables’.

One of the members suggests using Typeset for a preliminary literature review. Typeset can also explain math parts etc., and gives a summary of an intro or results, which is quite useful for understanding whether to invest your time in a certain paper or not. 

Another tool is Elicit.com, similar to the Typeset, which is used to get an idea of what’s written on a certain topic.

Another community scientist suggests using scienceOS. She uses the tool to ask scientific questions and get a summary of related research papers with references and sources. This software can also create a literature overview map to find foundational papers. There is also a chat to ask follow-up questions for refined answers, create tables to compare two answers and build diagrams for presentations. ‘It’s pretty cool that scienceOS is very open to feedback and implements new features regularly’, she adds.

Another tool that can be used to look for the connected papers is Connected Papers which also visualizes these connections in a graph.

Coding

Some community members share that they use ChatGPT to improve and simplify their work with R or Python. ‘I actually use ChatGPT a lot for coding. For me, this works really well. However, it will work better if you divide the tasks and try different prompts’, adds one of the mentees. ‘I also tried Phind for coding. I wouldn't say it was better than ChatCPT for the same task though. I use ChatGPT to write emails, read articles, generate ideas, and so on.

One of the scientists suggests using GitHub Copilot tool. ‘It is awesome for coding. It's free for students and "teachers", and it seems like they are quite lenient with the definition of the latter, - he says.

Writing

One of the community members working in scientific communication suggests using ChatGPT to help with writing. ‘I use this tool to simplify texts, find synonyms, and also adjust texts to, particularly targeted groups’, she says. ‘Of course, it is essential to do quality control of the output and use common sense when evaluating the results’. Some people said they use Wordtune for similar purposes.

‘Grammarly is a lifesaver to me as a non-native English speaker. I feel much more certain in the things I send away when I can check grammar in my texts’, says one of the researchers. Another member of the community suggests Trinka for the same purposes of proofreading and re-phrasing texts.

'Another great tool I recently found is the spell checker by websiteplanet', adds another scientist. 'I was amazed by the features: not only can you spot your mistakes, but you can also implement the edits it suggests automatically and just copy the fixed content. And it supports many languages'.

Preparing for an interview

‘I used science.OS when I was preparing for an interview to join a graduate school. With this tool I got an overview of the main publications by a professor, navigated through an overarching goal of their research, and therefore felt much more prepared for a discussion’, says a Sci.STEPS mentor.

Additional recommendations

Also you can find tons of AI recommendations on the channel of Dr. Andy Stapleton.

Have any tool to recommend to our community?

Let us know!

About Sci.STEPS

The Sci.Steps mentoring program aims to support scientists in every step they take in their career development. By connecting aspiring scientists with established professionals in their respective fields, this program facilitates an experience exchange and community building. Sci.STEPS provides targeted consultations for individuals, that need support in career orientation.