Artificial intelligence techniques have advantages and disadvantages when writing scientific research. What are your opinions about the new technology DeepSeek in this field?
As first opinion, after small use, I can say that it is powerful, helpful and coherent.
Don't use for any topics that involve China -- it has been too heavily censored to be useful in that context.
Scientists flock to DeepSeek: how they’re using the blockbuster AI model
Researchers are testing how well the open model can perform scientific tasks — in topics from mathematics to cognitive neuroscience...
"Scientists are flocking to DeepSeek-R1, a cheap and powerful artificial intelligence (AI) ‘reasoning’ model that sent the US stock market spiralling after it was released by a Chinese firm last week.
Researchers say DeepSeek has some notable advantages:
- The model is much cheaper to use than proprietary rivals, such as OpenAI’s o1. Using its online chatbot, DeepThink, is free.
- Researchers can download DeepSeek to their own servers and run and build on it, also for free, which is a plus for privacy.
- DeepSeek matched o1’s performance in preliminary tests on data-driven scientific tasks.
There also seem to be drawbacks:
- The model fails on some relatively simple tasks known also to trip up other large-language models, such as counting the number of US state names that contain the letter W.
- DeepSeek-R1 appears to be intermittently incapable of providing information about topics that are deemed to be sensitive by China’s government, although it’s unclear whether this is built into the model or is applied to its interface."
https://www.nature.com/articles/d41586-025-00275-0
Why scientists like the look of DeepSeek
"Like OpenAI’s industry-leading o1, DeepSeek-R1 is a ‘reasoning’ model that uses a ‘chain of thought’ method to solve complex tasks, including sometimes backtracking and evaluating its approach — making them more adept than earlier LLMs at solving scientific problems. But, unlike o1, DeepSeek has been released as ‘open-weight’, meaning that researchers can study and build on the algorithm. There are also mini ‘distilled’ versions of R1 to allow researchers with limited computing power to play with the model. And then there’s the price. Using DeepSeek’s interface, an experiment that cost more than US$300 with o1, costs less than $10 with R1, says AI scientist Mario Krenn."
https://www.nature.com/articles/d41586-025-00229-6
DeepSeek is really revolutionary, however, I don't think we should consider using it for paper writing or generating ideas. It's my personal belief AI is appropriate for debugging code or generating code (I use DeepSeek for this), however I would never under any circumstance let it do my work for me. I think of AI (DeepSeek included) as a tool only. Just as other kinds of AI, DeepSeek is not perfect and can actually make mistakes, so if we forget about ethics, I would still trust myself more to think outside of the box in research...
How are researchers using AI? Survey reveals pros and cons for science
Despite strong interest in using artificial intelligence to make research faster, easier and more accessible, researchers say they need more support to navigate its possibilities...
"A survey of nearly 5,000 researchers worldwide by publisher Wiley suggests that using artificial intelligence (AI) tools ( ChatGPT and DeepSeek) for processes such as preparing manuscripts, writing grant applications and peer review will become widely accepted within the next two years. Only 45% of the first wave of respondents (1,043 researchers) said that they had actually used AI to help with their research, and the most common uses they cited were translation, proofreading and editing manuscripts. However, the majority of those surveyed expressed interest in expanding their AI use..."
https://www.nature.com/articles/d41586-025-00343-5
OpenAI’s ‘deep research’ tool: is it useful for scientists?
The model produces cited, pages-long reports that might be helpful for generating literature reviews...
"Tech giant OpenAI has unveiled a pay-for-access tool called ‘deep research’, which synthesizes information from dozens or hundreds of websites into a cited report. The tool’s ability to write literature reviews and identify gaps in knowledge has impressed many scientists, but others have noted that its output is not up to the standard you’d expect from a human. Its launch follows that of a similar tool from Google, also called Deep Research, released in December. Artificial-intelligence systems like these could be used to update human-authored reviews more regularly than is feasible for humans, say experts, although OpenAI admits the tool is currently prone to hallucinations and false citations."
https://www.nature.com/articles/d41586-025-00377-9
DeepSeek is a solid tool for scientific research. It’s open-source and free, making it accessible to exploring new ideas. I find it especially useful for uncovering insights in complex datasets, which can spark innovative research directions.
However, while DeepSeek is a valuable resource, we don’t have to rely on it exclusively. It works best as part of a broader toolkit of methods and approaches. I appreciate the efforts of the DeepSeek team in developing this technology and believe that, with further improvements, it will continue to support scientific discovery. Ahmed Mounaf Mahdi
The AI cold war, tech disruption and the role of universities
DeepSeek’s meteoric, interdisciplinary success starkly highlights inefficiencies embedded within many American universities. While deep specialisation has been the engine behind countless scientific breakthroughs, it is increasingly misaligned with the cross-disciplinary nature of contemporary AI research...
https://www.universityworldnews.com/post.php?story=20250207090353961
- Badges
- Science topic
- Similar topics
- Medicine
- Counseling Psychology