I was reading an interesting section from the economist Noah Smith’s Substack blog Noahpinion. The section is about “How AI will change the science field, according to AI.” An academician, Kevin Bryan, asks Open AI’s Deep Research to write a paper about “the future uses of AI in quantitative social science research, and how those fields will have to change to accommodate AI.” Below is a summary of what AI said about this, which is a comprehensive and thoughtful list of its effects already:
Briefly, AI is transforming research. It helps generate ideas, summarize literature, and even draft papers. But it also raises concerns about accuracy, ethics, and overreliance.
How AI is Changing Research
Task | AI’s Role |
|---|---|
Writing Abstracts | AI creates summaries that even experts struggle to distinguish from real ones. |
Literature Reviews | AI condenses large papers into key points. |
Finding Sources | AI suggests relevant studies with citations. |
Brainstorming | AI generates hypotheses but needs human refinement. |
Risks & Challenges
Issue | Why It Matters |
Factual Errors | AI sometimes makes up facts and citations. – “fact checking” needed. |
Authorship Blur | Who really wrote the paper—AI or the researcher? |
Misconduct Risk | AI could be used to fake research data. |
Overreliance | Researchers might lose critical thinking skills. |
The Future of AI in Research
- By 2035, AI will be as common as spell checkers.
- Discipline-specific AI tools will emerge.
- Journals will use AI for submission screening and peer review.
- AI might generate research results with full analysis and write-ups.
My Own Takeaway
AI isn’t replacing researchers yet. If used wisely, it can enhance research. Considering the knowledge burden* for the academic scholars, humans will need intelligent machines to achieve scientific progress. But, misuse could weaken academic integrity. The future is inevitably AI and scholars working together, but misuse is a big question.
Source: https://open.substack.com/pub/noahpinion/p/at-least-five-interesting-things
* “Knowledge burden” – i.e. the knowledge frontier expands so rapidly that human cognition becomes one of the main limitations to identify which combinations of knowledge are likely to produce new useful ideas. – by Bloom, Jones, Van Reenen, and Webb, 2017.
Leave a Reply