CHEM 303: Laboratory in Biochemical Research (HC) Fall 2025

In general, gen AI models are are fundamentally different from a search engine because they "generate" original content rather than retrieving existing content.

Google search

Perplexity Prompt

Some gen AI models are more geared toward literature searching--i.e. finding scholarly journals--but most of those require a paid subscription. Most gen AI models provide summaries of information. 

NOTE: this is how I feel today. Gen AI is constantly evolving, and my thoughts may also evolve.

Possible acceptable uses

• Add to your existing body of knowledge
  • e.g. clarify unfamiliar words, concepts, or references from class meetings or readings
     
• Brainstorm ideas - research questions, syntheses, etc.
   
• Help gather vocabulary and search words around your topic
   
• Suggest a preliminary outline for your literature review
   
• Proofread text you've written
   
• Help understand unfamiliar concepts, but only with the intention of verifying against other reputable sources

Weaknesses/problems

• The average person does not know how to use gen AI effectively (esp. poor prompt writing)
   
• Ethical considerations
   
• Missing information
  • The corpus of information is limited; garbage in/garbage out and vice-versa
    • ChemRxiv (open) vs. Elsevier (paywalled)
    • Some publishers have licensed content to AI companies without telling authors who have signed away their rights in the publishing process (good for the corpus of information, but borders on unethical)
       
• Inaccurate information
  • If the AI model is trained on inaccurate data, it will output inaccurate data
  • Sources/references may simply be fake (AKA "hallucinations")
     
• Decline of critical thinking
  • i.e. doing things the easy way = little deep learning

Comparison of PubMed and Claude search

*Takeaways*

Exercise good judgement: just because you can use gen AI in every workflow doesn’t mean you should use it in every workflow.

Always cross-check information you get from gen AI against reliable sources, such as peer-reviewed journal articles

NIH guidance on the appropriate usage of artificial intelligence (AI) in the research application process (i.e. grant writing) (July 2025)

"NIH will not consider applications that are either substantially developed by AI, or contain sections substantially developed by AI, to be original ideas of applicants."

Why the new policy?

• AI use may result in plagiarism, fabricated citations, or other kinds of research misconduct. 
• AI makes it quicker (but not better!) to apply, and large numbers of applications from one PI burden the applicaiton review process.

The Use of Generative Artificial Intelligence Technologies is Prohibited for the NIH Peer Review Process (June 2023)

"The use of generative AI tools to output a peer reviewer critique on a specific grant application or contract proposal requires substantial and detailed information inputs. AI tools have no guarantee of where data are being sent, saved, viewed, or used in the future, and thus NIH is revising its Confidentiality Agreements for Peer Reviewers to clarify that reviewers are prohibited from using AI tools in analyzing and critiquing NIH grant applications and R&D contract proposals. Such actions violate NIH’s peer review confidentiality requirements."

Use of generative artificial intelligence technology in the NSF merit review process (December 2023)

• "NSF reviewers are prohibited from uploading any content from proposals, review information and related records to non-approved generative AI tools."
• "Proposers are encouraged to indicate in the project description the extent to which, if any, generative AI technology was used and how it was used to develop their proposal."

Why?

"Any information uploaded into generative AI tools not behind NSF's firewall is considered to be entering the public domain. As a result, NSF cannot preserve the confidentiality of that information."

EXAMPLE: ACS Publications

Artificial Intelligence (AI) Best Practices and Policies at ACS Publications

Article Submission

"All use of AI tools should be disclosed within the submission."

"The editor may, at their discretion, determine that the AI use in a given submission is too extensive, including (but not limited to) AI tools used to generate substantive commentary or extensive literature reviews."

Peer Review

"ACS Publications journals ask for the expert opinion of consulted reviewers and expect that those opinions are those solely of the reviewer....Additionally, third party services such as AI tools may store or use any information provided as a prompt to generate future text, which could result in a breach of the confidentiality of the peer review process.