Whether you’re engaged in an applied or pure research project at an undergraduate or graduate level, some substantial tasks demand a great deal of time and attention. Building a bibliography, summarizing a wide variety of existing research, drafting and rewriting, and ensuring data is thoroughly and appropriately analyzed is demanding work.
We can have our own digital assistants, but to be effective, we need different tools for different tasks. They can be “connected,” but this requires a level of programming skills most of us don’t have. However, if you’re diligent and conscientious, you can make a suite of tools work for you.
With about 30 new apps being released each day, AI is moving fast. Search There’s An AI For That to see what’s new and if it could be helpful to you.
Here are apps you can now use to:
1. Refine your research question or hypothesis
Before rushing into your research, spend time refining your research question or hypothesis so you can focus your energy on adding something to existing knowledge and understanding and push boundaries. Perplexity AI, which uses the Claude large language model, can really help with this task. One key feature is the material it produces is traceable, showing you the source it based its response on.
Simply asking, “What do we know about X?” will produce a very generic response. Better to ask, “What work exists that tests the idea that [state your question]. Show sources and highlight key research findings from 2020 to the present.”
2. Find relevant research
Research Rabbit uses a key paper you see as central to your research to find related papers by the same authors or authors who also found it a seminal trigger for their own work. In just seconds, a single paper submits leads to hundreds of related papers, which are mapped either by key content or by author. The tool is far more powerful than a Google Search with Google Scholar and much more effective. The papers you find yourself wanting to pursue are brought to you automatically. Papers written in languages other than English are translated on the fly.
3. Summarize available research
Having found the suite of material you think is relevant, you can save reading time by uploading PDFs, DOIs or URLs into Genei.io and ask it to summarize them. You can use the summaries to decide which paper or papers to really dig into.
Better still, with Tactic, you can submit a range of papers and ask it to summarize responses to a question. I will do so for each paper. For example, using 10 papers on positive psychological capital, a researcher asked “What are the key findings from each paper and how do the findings differ across these studies?” Tactic summarized each and did a compare and contrast — something a Bing summary can’t do.
Videos can also be summarized, including podcasts, TED Talks and YouTube videos. Share the video with Summarize.tech and you get a short written summary. Look at the summary of Susskind’s physics lecture as an example of what this powerful system can do.
4. Analyze data
Statistical data analysis used to be a complex task, especially when all the data was held in a box of punch cards or on unique data tapes. Now with Excel, Sheets and other spreadsheet applications, we can automate data analysis. Rather than writing out formulae, we can use natural language to interrogate small or large datasets. There are several options available, but Deepsheet and InsightJini or ChatGPT advanced data analysis work well for this task.
Social scientists undertake a lot of interviews and seek to analyze the transcripts using coding systems and sentiment analysis. Automating transcription speeds this process and reduces time and costs on this task. Transcript.LOL and Scribewave do a solid job of transcribing audio recordings. Researchers can then run these transcripts through Sentigem, which highlights key points that suggest an attitude, value or observation, making coding easier. NVivo, which is a qualitative data analysis software program, is also popular. The electronic process of coding provided by NVivo allows for efficient and effective coding that helps in generating relationships and allows for easier retrieval of similar ideas and themes. Although not strictly an AI tool, it is making use of algorithms to analyze text.
5. Write up the research
Most research documents — papers, journal articles, a thesis — go through several drafts. Version control on these drafts is a key challenge. More than once a student has submitted the wrong version and then tried to retrieve it only to discover the paperwas graded.
Supervisors share a concern about the quality of writing. All writing can be improved. Researchers who use writing assistants such as Grammarly or Hemmingway can improve not just their grammar, but also the quality of their writing by turning on the AI assistants. Alternatively, cutting and pasting text into ChatGPT and asking it to “correct and improve this writing and explain what changes have been made and why” can produce helpful responses.
There is a growing list of AI-enabled writing assistants that can also be helpful. Microsoft’s Co-Pilot 365, which can be added to Word for a monthly fee of US$30 per user, can perform many of the writing support tasks researchers need.
6. Present the research
You may have to present your work to others at seminars, workshops, conferences, symposiums and keynote addresses. Great graphs from data and powerful animated presentations can make a difference in how your peers receive your work. Several AI tools can be used to generate effective and focused presentations automatically from plain text. beautiful.ai, SlidesAI and Prezo all do good work.
You can make even more impact by including short video presentations, using animated characters (or historical ones) to summarize key findings or ideas. Text-to-video production is also very efficient, especially when you let Deepbrain.ai or Taleblocks do the work with you. Deepbrain, which uses avatars to present your work and ideas, can lead to impactful presentations.
Good graphics from data are also important in telling a story. Chat2CSV takes data from a spreadsheet, allowing you to generate visual representations of the material using natural language commands.
Make sure to adhere to all institutional policies
You must keep in mind your institutional policies with respect to the ethical and academically appropriate use of AI tools. You don’t want to be accused of academic misconduct. Make sure you follow the appropriate format for referencing your use of AI in your work.