Human vs. machine: Neil Spooner takes on ChatGPT

Written by ChatGPT, Neil Spooner

In this unusual interview feature, Editor in Chief of Bioanalysis, Neil Spooner, provides his commentary alongside an interview with ChatGPT. Although generally impressed by ChatGPT’s comprehensive answers, Neil shares his own take on the questions and highlights where the intricacies of the bioanalytical field may have been overlooked.


ChatGPT
Artificial Intelligence Chatbot
OpenAI (CA, USA)

ChatGPT is an AI-powered language model created by OpenAI and is a sibling model to InstructGPT. The model was trained using Reinforcement Learning from Human Feedback. ChatGPT is designed to respond to text-based queries and can learn patterns in language and generate text analogous to human phrasing. As part of the broader field of AI known as natural language processing, ChatGPT has been taught to understand and interpret human language.

 

Neil Spooner
Founder and Director
Spooner Bioanalytical Solutions (Hertford, UK)

Neil Spooner (PhD, C.Chem, FRSC) is the Founder and Director of Spooner Bioanalytical Solutions, a consultancy helping companies to integrate microsampling and patient centric sampling of biological fluids into pre-clinical, clinical and bioanalytical workflows. Neil is also the Chair of the Patient Centric Sampling (CIC) interest group, a Senior Visiting Research Fellow at the School of Life and Medical Sciences, University of Hertfordshire (UK), the Editor in Chief of the Bioanalysis Journal and the Co-Chair of the Reid Bioanalytical Forum.

 

Questions

How do I become a bioanalyst?

ChatGPT: To become a bioanalyst, you typically need to follow these steps:

  1. Obtain a Bachelor’s degree in biology, biochemistry, biotechnology, or a related field. Some bioanalytical jobs may require a higher level of education, such as a Master’s or Doctoral degree.
  2. Gain hands-on laboratory experience through internships, research projects, or work experience to develop practical skills in laboratory techniques and instrumentation.
  3. Become familiar with the regulatory requirements for the industry, such as Good Laboratory Practices (GLP) and Good Manufacturing Practices (GMP).
  4. Acquire specialized training or certifications in areas, such as mass spectrometry, chromatography, or immunoassays.
  5. Attend conferences, workshops and industry events to build a network of professionals in the field. Join professional organizations such as the American Association of Bioanalysts (AAB; MO, USA) or the Society for Laboratory Automation and Screening (SLAS; IL, USA).
  6. Look for bioanalytical positions in academic research institutions, pharmaceutical/biotechnology companies, government agencies, or contract research organizations (CROs).
  7. Stay up-to-date with the latest advancements in technology and techniques in the field and continue to develop your skills through training and professional development opportunities.

It’s important to note that the requirements for becoming a bioanalyst may vary.

Neil: This is a great answer and seems to cover all the bases that I would have recommended myself. ChatGPT seems to have a bias towards conferences for clinical chemists rather than those performing bioanalysis for pharmaceutical drug development. Maybe this tells us a lot about how the pharmaceutical bioanalytical community markets itself! Also, it does not have any awareness that some of the suggestions such as attending conferences are out of the reach of even experienced bioanalysts due to cost constraints. It would also have been helpful if ChatGPT had made follow up suggestions for some of the activities to help those interested to progress. For example, which Universities are more likely to lead to jobs in bioanalysis; which organizations might offer internships, work experience opportunities, etc.; where might you get specialized training; which organizations are currently offering positions. Maybe we just need to ask it these follow-up questions.

What are the benefits and downsides of being a bioanalyst?

ChatGPT: Benefits of being a bioanalyst:

  1. Due to the increasing importance of biotechnology and biopharmaceuticals in healthcare and other industries, bioanalysts are in high demand.
  2. Bioanalysts can work in various fields, such as biotechnology, pharmaceuticals, environmental science and food science.
  3. Work can be intellectually stimulating and rewarding as bioanalysts work on a range of projects that involve analyzing biological samples, developing and validating methods and interpreting data.
  4. Bioanalysts often work in teams with other scientists, engineers and professionals, which can provide opportunities for collaboration and networking.
  5. There are numerous opportunities for bioanalysts to grow their careers by obtaining advanced degrees, pursuing certifications and gaining experience in different areas of the field.

Downsides of being a bioanalyst:

  1. Performing repetitive tasks such as sample preparation and data entry can be monotonous.
  2. Bioanalysts may be under pressure to meet tight deadlines for projects, which can be stressful.
  3. Technical challenges during experiments, such as contamination or equipment malfunctions, can be frustrating and time-consuming.
  4. Hazardous materials such as chemicals, biological agents or radiation can pose safety risks for bioanalysts if proper precautions are not taken.
  5. Bioanalysts may work long hours, including weekends and holidays, to meet project deadlines or respond to urgent issues.

Neil: Again, this is a solid, if predictable, answer. I would personally have given additional answers around feeling you are contributing to the development of new medicines to help human health and well-being. Also, the broad exposure to many aspects of drug development that is obtained by bioanalysts gives them great opportunities to develop their career along other pathways.

What is the ideal approach for performing quantitative bioanalysis, immunoassay, or LC-MS?

ChatGPT: There is not one ‘ideal’ approach for performing quantitative bioanalysis as both immunoassay and LC-MS have their advantages and limitations and the choice of method depends on several factors such as the analyte of interest, sample matrix, required sensitivity and selectivity and regulatory requirements.

Immunoassay is a widely used method for quantitative bioanalysis, especially for large molecules, such as proteins, antibodies and peptides. Immunoassay methods are highly specific and sensitive, can detect low levels of analytes and are relatively simple to perform. However, immunoassays are subject to matrix interference and cross-reactivity and their performance can vary depending on the quality of the antibodies and reagents used.

LC-MS is another widely used method for quantitative bioanalysis, especially for small molecules such as drugs, metabolites and lipids. LC-MS methods offer high selectivity, sensitivity and specificity and can detect a wide range of analytes in complex matrices. However, LC-MS methods require specialized equipment, expertise and sample preparation and are subject to matrix effects, ion suppression, etc.

Some analytes may be better suited for immunoassay, while others may require LC-MS. In some cases, a combination of both methods may be necessary to achieve the desired sensitivity and selectivity. The choice of method should be based on a careful evaluation of the advantages and limitations of each approach and the specific needs of the bioanalytical project.

Neil: This is a comprehensive answer to a complicated question and ChatGPT gets it mostly right. I was particularly impressed with it mentioning that a combination of these approaches might be needed in some cases. However, the statement that LC-MS needs specialized equipment and expertise is not totally correct. In my opinion, both approaches require this. ChatGPT has also missed a key consideration when using these analytical approaches in clinical drug development, particularly for proteins. LC-MS methods are quicker to develop but potentially have higher overheads long term and can, therefore, be more expensive if long term support is required. Whereas, immunoassay approaches tend to take longer to develop as production of reagents is slow, but long-term running costs tend to be more affordable. Another important factor that ChatGPT missed is that the two techniques can give different answers when measuring the same analyte in the same sample. This is due to differences in specificity in that they are not actually measuring the exact same entity. Both answers are correct, just different.

Which areas of bioanalysis should we be automating and which are best left to human effort?

ChatGPT: Automation can bring significant benefits to bioanalysis, such as improving efficiency, consistency and reproducibility, reducing the risk of error and increasing throughput. However, not all aspects of bioanalysis can be effectively automated and there are some tasks best left to human efforts. Here are some examples:

Areas of bioanalysis that can be effectively automated:

  1. Automation can be particularly effective for repetitive and labor-intensive tasks, such as sample handling, extraction and clean-up.
  2. Accurate and precise liquid handling is critical for many bioanalytical methods, particularly those involving multiple reagents and samples.
  3. Used in many bioanalytical methods, chromatography is a key separation technique. Automated chromatography systems can help improve reproducibility and reduce the risk of error, whilst also increasing throughput, enabling the analysis of large numbers of samples.
  4. A powerful and widely-used analytical technique, mass spectrometry, can be complex and time-consuming to set up and optimize.

Areas of bioanalysis that are best left to human efforts:

  1. Whilst automation can be useful for data processing and interpretation, the final analysis and interpretation of the data should be left to human experts who can apply their knowledge and expertise to the results.
  2. When problems arise in bioanalytical methods, it is often necessary to have an experienced scientist to identify and solve the problem. Troubleshooting requires creativity, critical thinking and problem-solving skills that cannot be easily automated.
  3. Developing and optimizing bioanalytical methods requires a deep understanding of the underlying principles and a high degree of technical expertise.

Neil: I found this to be the weakest of the responses from ChatGPT in terms of how it might compare to how I would have answered the question. It seems to have gone off on a tangent. For example, I do not think that accurate and precise liquid handling is one of the key aspects when considering automation versus human work. Yes, it needs to tick this box, but it is not necessarily important that automated systems do this better than humans. For automated chromatography systems, the ChatGPT answer would be better if it had stated that this applies to method development. One particularly interesting aspect is that as an AI tool itself, ChatGPT did not consider how AI might contribute to the areas which it considered are best left to humans, such as data interpretation, problem solving and optimizing methods. An automated system that suggests to a human where they might prioritize their efforts and thoughts would surely be better than just relying on whether the right tech expert happens to be available in the lab on a particular day to sort out issues.

Concluding remarks from Neil

I was blown away when I read these responses from ChatGPT and I have been excitedly – and annoyingly – sharing with family and friends ever since! One thing that struck me was not only the valuable content it has pulled together in a concise manner but also the language used. I have read many draft manuscripts and reports and few are as well written as what ChatGPT has delivered. However, on further thought, whilst I am impressed with what ChatGPT has come up with, I think it is fair to say that the responses are probably quite bland and predictable. There is no passion or intuition. No novel ideas and approaches. Hence, it feels like the ChatGPT tool is just that at the moment. A tool for pulling together information in a coherent manner, that in some cases is more user friendly than the output from a typical current web search tool. I am very excited to learn more about this tool and to see how it develops. I can definitely see a major place for tools like this in the information rich world we live in.