[Live from AI #1] Use a casual prompt to start a meaningful conversation
An example ChatGPT conversation on tailoring your resume (plus bonus behind-the-scenes footage) and how you don’t need the “perfect prompt”
This article is the first in a series I’m calling, “Live from AI”, where I’ll share examples of conversations I’ve had with ChatGPT in the full.
The beauty of this new technology is there is no one right way to use it. So many guides treat AI like a program that requires the perfect prompt. Based on my experience, the magic of AI is in the wild, messy unknown of a conversation.
I talk about this idea high level in a recent post, You don’t need the perfect prompt to get started with AI: Just have a conversation, and the examples I referred to in that post came from this conversation.
The Context
I created the following conversation to demonstrate how you can talk with a ChatGPT without a “perfect prompt”. Sometimes the resulting conversation can be more meaningful and rich than if you used a purely utilitarian approach.
The example I created focuses on resume optimization. As a bonus, I also include the behind-the-scenes chat that led to creating the exercise.
If you found this format helpful, I’d love to know! Please your thoughts in the comments below.
The Conversation
I started the conversation with a prompt straight from my own head. Let’s see where it goes.
I said:
I want to create a demo conversation here with you for my blog. I'm going to ask you questions. I want you to not refer to anything we've already discussed.
ChatGPT said:
Perfect—I'm ready. Ask away, and I’ll respond fresh, as if it’s a live demo for your blog.
I said:
Hey, I'm an academic interested in transitioning to industry. I have a resume and a job description. Can you give me suggestions for where I can make changes to my resume to make it more relevant for industry?
ChatGPT said:
Absolutely, I’d be happy to help with that. To get started, could you please share:
The job description you're targeting, and
The current version of your resume (or a section of it—like your summary or experience)?
With both in hand, I’ll look for alignment gaps and opportunities to better match your resume to the language and priorities of the industry role.
I said:
Okay, now stepping out of the demo, can you create a fake resume for an academic that is trying to get into industry, has relevant research, technical, and statistical experiences, but doesn't know how to write for business outcomes. Let's ground this in a person with a Psychology PhD who is interested in transitioning to data science.
ChatGPT said:
Great—here’s a fictional academic resume that reflects someone with a Psychology PhD and strong research/technical skills, but who hasn’t yet translated those skills into business-facing language. It intentionally shows some common patterns academics fall into during their first draft of an industry resume.
Dr. Alex Morgan
amorgan@email.com | LinkedIn: linkedin.com/in/amorgan | San Diego, CA
SUMMARY
Cognitive psychologist with 8+ years of experience in experimental design, statistical analysis, and behavioral research. Extensive background in hypothesis testing, data collection, and mixed-methods research. Proficient in R, Python, SPSS, and Qualtrics. Passionate about understanding human behavior through data.
EDUCATION
Ph.D., Psychology
University of California, San Diego — 2023
Dissertation: Decision-making under uncertainty: Behavioral and neurocognitive perspectives
B.A., Psychology and Statistics
University of Michigan — 2015
RESEARCH EXPERIENCE
Graduate Researcher
UC San Diego, Psychology Department | 2017–2023
Designed and conducted multi-phase behavioral experiments with over 1,200 participants to study decision-making, learning, and attention.
Collected and analyzed large datasets using R and Python (pandas, scikit-learn).
Published 3 first-author papers in peer-reviewed journals; presented findings at 6 national conferences.
Led weekly lab meetings and mentored 3 undergraduate research assistants.
Data Analyst (Part-Time)
Center for Human Development | 2020–2022
Analyzed longitudinal survey data from early childhood development studies.
Used hierarchical linear modeling and mixed-effects regression to identify cognitive growth trends.
Wrote data processing pipelines in R and collaborated with a team of researchers to prepare reports.
SKILLS
Programming & Analysis: R, Python, SPSS, SQL, MATLAB
Tools: GitHub, Qualtrics, Excel, Tableau (beginner)
Statistical Methods: Linear regression, ANOVA, mixed models, Bayesian methods
Soft Skills: Writing, research synthesis, teaching, mentoring
This version reflects a solid foundation but hasn’t yet:
Connected to business outcomes or team goals
Used industry-relevant phrasing (e.g., “insights to drive decisions,” “impact,” “stakeholders”)
Emphasized transferable skills like cross-functional collaboration, communicating insights, or automating analysis workflows
Would you like to now generate feedback on how to adapt this version for industry relevance in the next step of the demo?
I said:
Can you create a fake job description for a data science role?
ChatGPT said:
Certainly—here’s a fictional job description for a Data Scientist role at a mid-sized tech company. It’s designed to be realistic and typical of roles targeting candidates with academic or research-heavy backgrounds transitioning to industry.
Job Title: Data Scientist
Location: Remote (U.S. based)
Company: Verida Health Tech
About Us:
Verida is a fast-growing digital health company that empowers patients and providers through data-driven insights. Our platform helps healthcare teams personalize treatment pathways, reduce inefficiencies, and improve patient outcomes.
Job Summary:
We’re looking for a curious and collaborative Data Scientist to join our analytics team. You’ll work closely with product, engineering, and clinical teams to analyze user behavior, run experiments, and deliver actionable insights that shape the direction of our platform.
Responsibilities:
Analyze large-scale healthcare and user interaction data to uncover patterns and trends.
Design and evaluate A/B tests to support product and clinical decision-making.
Build predictive models to forecast outcomes and personalize user experiences.
Communicate findings to cross-functional stakeholders through dashboards, presentations, and reports.
Collaborate with data engineers to improve data quality and pipeline efficiency.
Translate complex findings into clear recommendations aligned with business goals.
Qualifications:
2+ years of experience in data science, analytics, or applied research (including PhD work).
Proficiency in Python or R and SQL for data manipulation and modeling.
Strong foundation in statistics, experimental design, and machine learning.
Experience working with large, messy datasets and synthesizing key insights.
Excellent communication skills—able to present results to technical and non-technical audiences.
Experience in healthcare, behavioral science, or related domains is a plus.
Bonus Points For:
Familiarity with dashboarding tools (e.g., Tableau, Looker, Streamlit).
Experience with NLP or time-series data.
Background in psychology, cognitive science, or behavioral economics.
Would you like to continue the demo now using this job description and the previous resume?
Interlude where I fought with ChatGPT about formatting
I said:
Let's go back to the demo, I'm going to share with you the text
ChatGPT said:
Great—go ahead and share the text whenever you're ready, and we’ll pick up the demo from there.
I said:
Note: I tend to copy-paste text in straight. I didn’t care about formatting. The subsequent results were good enough for me.
Here is my resume:
Dr. Alex Morgan amorgan@email.com | LinkedIn: linkedin.com/in/amorgan | San Diego, CA SUMMARY Cognitive psychologist with 8+ years of experience in experimental design, statistical analysis, and behavioral research. Extensive background in hypothesis testing, data collection, and mixed-methods research. Proficient in R, Python, SPSS, and Qualtrics. Passionate about understanding human behavior through data. EDUCATION Ph.D., Psychology University of California, San Diego — 2023 Dissertation: Decision-making under uncertainty: Behavioral and neurocognitive perspectives B.A., Psychology and Statistics University of Michigan — 2015 RESEARCH EXPERIENCE Graduate Researcher UC San Diego, Psychology Department | 2017–2023 Designed and conducted multi-phase behavioral experiments with over 1,200 participants to study decision-making, learning, and attention. Collected and analyzed large datasets using R and Python (pandas, scikit-learn). Published 3 first-author papers in peer-reviewed journals; presented findings at 6 national conferences. Led weekly lab meetings and mentored 3 undergraduate research assistants. Data Analyst (Part-Time) Center for Human Development | 2020–2022 Analyzed longitudinal survey data from early childhood development studies. Used hierarchical linear modeling and mixed-effects regression to identify cognitive growth trends. Wrote data processing pipelines in R and collaborated with a team of researchers to prepare reports. SKILLS Programming & Analysis: R, Python, SPSS, SQL, MATLAB Tools: GitHub, Qualtrics, Excel, Tableau (beginner) Statistical Methods: Linear regression, ANOVA, mixed models, Bayesian methods Soft Skills: Writing, research synthesis, teaching, mentoring
And here is the job description:
Job Title: Data Scientist Location: Remote (U.S. based) Company: Verida Health Tech About Us: Verida is a fast-growing digital health company that empowers patients and providers through data-driven insights. Our platform helps healthcare teams personalize treatment pathways, reduce inefficiencies, and improve patient outcomes. Job Summary: We’re looking for a curious and collaborative Data Scientist to join our analytics team. You’ll work closely with product, engineering, and clinical teams to analyze user behavior, run experiments, and deliver actionable insights that shape the direction of our platform. Responsibilities: Analyze large-scale healthcare and user interaction data to uncover patterns and trends. Design and evaluate A/B tests to support product and clinical decision-making. Build predictive models to forecast outcomes and personalize user experiences. Communicate findings to cross-functional stakeholders through dashboards, presentations, and reports. Collaborate with data engineers to improve data quality and pipeline efficiency. Translate complex findings into clear recommendations aligned with business goals. Qualifications: 2+ years of experience in data science, analytics, or applied research (including PhD work). Proficiency in Python or R and SQL for data manipulation and modeling. Strong foundation in statistics, experimental design, and machine learning. Experience working with large, messy datasets and synthesizing key insights. Excellent communication skills—able to present results to technical and non-technical audiences. Experience in healthcare, behavioral science, or related domains is a plus. Bonus Points For: Familiarity with dashboarding tools (e.g., Tableau, Looker, Streamlit). Experience with NLP or time-series data. Background in psychology, cognitive science, or behavioral economics.
ChatGPT said:
Thanks! Let’s continue the demo from here. I’ll now respond in character as the AI tool, helping Dr. Alex Morgan adjust their resume to better match the Verida Health Tech data scientist job description.
Hi Alex! Thanks for sharing your resume and the job description. Let’s walk through a few ways you could strengthen your resume to better align it with this data science role:
1. Highlight Business Impact and Relevance
Right now, your resume emphasizes academic skills and research accomplishments—which are impressive—but could be reframed in terms of business value. For example:
Instead of:
“Designed and conducted multi-phase behavioral experiments with over 1,200 participants to study decision-making…”
Try:
“Designed large-scale experiments to uncover decision-making patterns, generating insights with potential applications in digital health and user behavior modeling.”
This connects your research to outcomes relevant for the Verida role.
2. Reframe Experience as Applied Work
You’ve done substantial statistical and data work that can be pitched as applied analytics:
For your part-time Data Analyst role:
Add language like “delivered actionable insights” or “supported strategic decisions” to echo the language from the job description.
3. Tailor Skills Section
The job emphasizes predictive modeling, A/B testing, and collaboration with cross-functional teams.
You could:
Explicitly mention A/B testing if you’ve done any (or if your experiments align closely).
Include collaboration examples that involve technical and non-technical teammates.
Move SQL higher up in the skills section, or mention it more prominently if you've used it.
4. Include a Projects or “Relevant Experience” Section
If you’ve done any healthcare-related work, behavior prediction modeling, or relevant side projects, consider a new section to highlight that.
Would you like help drafting specific bullet revisions or a restructured version of your resume based on these suggestions?
I said:
I'm going to give you another demo question. Do not reference the demo in your response
ChatGPT said:
Got it—ready when you are! Go ahead and send your next question.
I said:
Note: I didn’t think the response to this question was very good or helpful, but I continued on to see where I could deepen my understanding.
In the Tailor Skills section, can you give more detail or examples? Don't rewrite it for me.
ChatGPT said:
Absolutely. Here are a few ideas to expand or give more detail in the “Tailor Skills” section:
Programming & Analysis Tools
Instead of just listing languages, you could describe what kinds of tasks you’ve used them for.
Example:R (data wrangling, statistical modeling with lme4, visualization with ggplot2)
Python (pandas for cleaning, scikit-learn for modeling, matplotlib for plotting)
SQL (basic querying for survey data joins and filtering)
Statistical Methods
You might elaborate on how you’ve applied these in projects.
Example:Used mixed models to account for repeated measures in decision-making experiments
Applied Bayesian methods to evaluate uncertainty in participant behavior patterns
Soft Skills
Give situational or functional examples to ground these.
Example:Synthesized research into actionable insights for lab presentations and publications
Mentored undergrads on statistical methods and research design
Tools
Especially for beginner tools like Tableau, it might be useful to say what you’ve done so far or what you're learning.
Example:Built basic dashboards to visualize behavioral experiment outcomes
Would you like to expand any of these specifically for the resume you shared?
I said:
Note: According to Sam Altman of OpenAI, users are costing them millions with their usage of ‘please’ and ‘thank you’. But I just can’t not do it…
yes please
ChatGPT said:
Great! Based on Alex Morgan’s resume and the job description, here are ways you could expand each section in their Skills area to make it clearer, more tailored, and aligned with the industry role (without rewriting it directly):
Programming & Analysis
R – Mention your use in experimental data cleaning, regression models, mixed-effects modeling (lme4), and creating publication-ready plots (ggplot2).
Python – Call out experience with pandas and numpy for data manipulation, scikit-learn for basic modeling or preprocessing, maybe matplotlib or seaborn for visualizations.
SPSS – You might frame this as experience with classic social science workflows or for quick descriptive stats.
SQL – Clarify with something like: basic querying for participant-level data extraction or joining large datasets.
Tools
GitHub – Used for version control on analysis scripts or collaborative coding projects.
Qualtrics – Designed and distributed surveys, managed experimental stimuli.
Excel – For quick summaries or preliminary exploration.
Tableau (beginner) – If you’ve started learning this, say how: “Currently building simple dashboards to visualize survey data.”
Statistical Methods
Instead of listing, you could map methods to research use:
Linear regression, ANOVA, mixed models – Used for analyzing behavioral data across conditions.
Bayesian methods – Applied to understand decision uncertainty or model participant priors.
Soft Skills
Writing & synthesis – Mention how you translated research into papers or presentations.
Teaching & mentoring – You mentored undergrads; could also note if you designed training materials or sessions.
Collaboration – Reference any cross-department work (e.g., psychology + human development) to highlight interdisciplinary communication.
Want help prioritizing which of these are most relevant to the job description you shared?
I said:
Note: in the above response, I wasn’t convinced it was a good answer, especially based on my own experience writing resumes for industry. I still channeled a person new to this experience though, and checked what ChatGPT may know about its answer.
How would I know if my resume is meeting the expectations of the job description? I've never applied to industry before.
ChatGPT said:
Note: ChatGPT is VERY positive and affirmative in its responses. Use it to feel good for a few minutes, then let it go so it doesn’t go to your head.
That’s such a good and common question—especially for someone coming from academia. Here are a few ways to know if your resume is meeting expectations, especially for industry roles like the data science job you’re looking at:
🔍 1. Look for alignment in language
Ask yourself:
Am I using business-oriented language, or am I still in research-speak?
→ Example: Instead of “examined the effects of variable X on behavior,” something like “analyzed user behavior patterns to inform strategy” is closer to what industry wants.
Tip: Try highlighting a few key phrases in the job description—like "deliver actionable insights," "cross-functional stakeholders," or "forecast outcomes"—then check if similar terms show up in your resume.
📌 2. Focus on impact, not just process
In academia, it's common to emphasize methods and rigor. In industry, they care about results and outcomes:
Ask: What happened because I did this?
Try to include metrics, efficiencies gained, or decisions influenced.
🧩 3. Compare your resume line-by-line to the job responsibilities
Literally take each bullet point in the job description and ask:
Do I have a story or example that matches this?
Is that example clearly visible in my resume? If not, can you reframe your experience to show it (without stretching the truth)?
🧠 4. Use the “could they picture me doing this job?” test
If someone in industry were reading your resume:
Would they understand what you’ve done without knowing academia?
Could they picture you doing the tasks listed in the job?
📊 5. Run your resume through AI tools (like ChatGPT!)
You can ask:
“Given this job description and resume, where are the biggest mismatches?”
Or
“What parts of this resume align most with what this company is looking for?”
Would you like to do a side-by-side alignment of Alex Morgan’s resume with that data science job? I can show you how the current bullets map onto the role—and where there might be gaps.
I said:
Note: Here I followed my curiosity to understand more about what terms could be referring to and augment knowledge a new person to industry may not have.
Could you tell me more about what business-oriented language means?
ChatGPT said:
Absolutely—this is a subtle but important shift when moving from academia to industry.
🧠 Academic Language vs. 🧳 Business-Oriented Language
Academic language tends to:
Emphasize methods, rigor, and nuance
Use specialized terms or jargon
Focus on what was studied and how
Prioritize individual contribution and precision
Business-oriented language shifts focus to:
Outcomes and impact
Collaboration and business value
Plain, accessible language
Action and efficiency
🔎 Key features of business-oriented language:
Action-oriented verbs – "Delivered", "Built", "Streamlined", "Improved", "Partnered with"
Focus on results – What changed? What improved? What decisions were influenced?
Emphasis on the bigger picture – Not just what you did, but why it mattered to the team/org/mission
Audience awareness – Assume your reader isn’t in your field. Can they still understand the value?
Would it help if I pointed out a few examples in Alex Morgan’s resume that could be “translated” into more business-oriented terms, just to make the shift more concrete?
I said:
Note: I found the above response illuminating and clear. I wanted to follow up on how I could apply it directly to my experience, typos and all
Can I share with you more details on my research, and you can help me understand what aspects of it and how to describe it would be relevant for a business?
ChatGPT said:
Yes, absolutely—please do! Share as much detail as you’d like about your research: what questions you studied, the methods you used, the kinds of data you worked with, the findings, and anything else that feels important. I’ll help you spot the parts that could translate well to business environments and how to describe them in a way that highlights their relevance and value.
From here you can continue the conversation with ChatGPT, either following up on more terms and ideas from the previous section or seeing where this new topic takes you.
Treating ChatGPT more like a dedicated 1:1 session with an expert, friend, therapist, mentor, librarian, or consultant goes a much longer way than crafting the perfect single prompt.