Discover more from Data-driven VC

Where venture capital and data intersect. Every week. Join 13,600+ thought leaders from VCs like a16z, Accel, Index, Sequoia, and more.

Your Personal 10x Analyst: Everything You Need to Know About ChatGPT Code Interpreter

DDVC #44: Where venture capital and data intersect. Every week.

Andre Retterath

Jul 13, 2023

👋 Hi, I’m Andre and welcome to my weekly newsletter, Data-driven VC. Every Thursday I cover hands-on insights into data-driven innovation in venture capital and connect the dots between the latest research, reviews of novel tools and datasets, deep dives into various VC tech stacks, interviews with experts, and the implications for all stakeholders. Follow along to understand how data-driven approaches change the game, why it matters, and what it means for you.

Current subscribers: 11,345, +192 since last week

Brought to you by PredictLeads - Company-level datasets for Data-Driven VCs

PredictLeads works with top Silicon Valley VCs and provides them with structured News Data, Jobs Data, Technologies, Key Customers Data, and more.
Data-driven VCs use the data to identify net-new opportunities and to track known startups better. Contact us to get a FREE report on 100 startups or sign up and try out our API.

Test for free

Driven by my profound fascination for advancements in AI and my continuous quest for boosting productivity, I devote this article to a rundown of my latest explorations with OpenAI's novel Code Interpreter capability in ChatGPT. I delve into the specifics of this feature and its prospective implications for knowledge professionals across various fields. Additionally, I offer a thorough step-by-step walkthrough to help you jumpstart your own journey with this innovative tool. Enjoy!

What Is ChatGPT Code Interpreter?

Code Interpreter is the newest addition to OpenAI's ChatGPT, particularly with the GPT-4 model, enabling the execution of Python code live within a secure sandbox. While it may initially seem like a feature exclusively designed for programmers, it can actually assist a wide range of users with numerous tasks such as:

Data Analysis: Users can perform various data analyses such as statistical analysis, exploratory data analysis, hypothesis testing, and predictive modeling.
Data Visualization: Users can create various types of charts and plots to understand and present their data better. This includes bar graphs, scatter plots, heat maps, line graphs, pie charts, and even complex 3D visualizations.
Machine Learning: Users can build, train, and evaluate machine learning models right within the chat. They can also use it to explain and interpret the results of these models.
File Manipulation: Code Interpreter can be used to read and write different types of files such as TXT, PDF, DOC, DOCX, JPEG, PNG, MP4, AVI, CSV, JSON, XML, XLS, XLSX, CPP, PY, HTML, PDF, DB, SQLite. Moreover, it can be used to convert file formats or perform operations on the data within these files.
Financial Modeling: Financial analysts can use the tool to analyze financial data, build financial models, and make forecasts.
Teaching and Learning: It can be used as a teaching tool to help students learn programming, data analysis, machine learning, and more. Individuals can also use it for self-study.
Automation of Tasks: The Code Interpreter can be used to automate repetitive tasks. For example, one can write a script to clean data, automate data entry tasks, leverage OCR to extract information from a PDF or schedule certain operations.
Prototyping and Testing: Developers can use Code Interpreter to quickly write and test small pieces of code, experiment with new libraries or functions, and debug their code.
Web Scraping: Although this feature may be limited due to the sandbox environment, users can still write code to parse and analyze pre-downloaded HTML or XML data.
Text Processing and NLP: Users can perform text processing tasks, sentiment analysis, text classification, and other natural language processing tasks.
….

These are just a few examples, and the possibilities are almost endless. Code Interpreter essentially brings the power of Python and its extensive libraries (access over 300 packages, see overview here) into the chat environment, allowing users to interact with code in a more natural and intuitive way.

Potential Implications for Knowledge Workers

The introduction of Code Interpreter in ChatGPT has several implications for knowledge workers, such as data analysts, researchers, investment professionals, consultants, and others who rely heavily on information processing and decision-making.

In line with the mantra of this newsletter, I’m personally most excited about its potential impact on the effectiveness, efficiency, and inclusiveness of the VC investment process. Having played around with the tool for a few days, I expect the following opportunities (pros):

Increased EFFICIENCY: Code Interpreter can automate many tasks that would otherwise be time-consuming, freeing up time and enabling knowledge workers to focus on more complex problems that require human judgment.
Improved Accessibility/INCLUSIVENESS: The feature makes programming and data analysis more accessible to individuals who may not have a strong background in these areas. This can democratize access to these skills and allow more people to leverage the power of data in their work.
Enhanced Decision-Making/EFFECTIVENESS: By providing the ability to quickly analyze and visualize data, Code Interpreter can aid in decision-making. Knowledge workers can use real-time data and models to inform their decisions, improving accuracy and outcomes.
Promotion of Lifelong Learning: As an interactive tool, Code Interpreter can be an effective learning platform. Knowledge workers can use it to continuously upgrade their skills, learn new programming techniques, and stay updated with the latest in AI and machine learning.
Innovation and Creativity: With the ability to experiment with code and a variety of data, knowledge workers can be more innovative and creative. They can quickly test out new ideas, analyze results, and iterate, driving innovation in their fields.

While it certainly unlocks a wide spectrum of opportunities, it also comes with some drawbacks (cons):

Risk of Overreliance: Although AI tools like the Code Interpreter are powerful, they can sometimes produce incorrect or misleading results. Therefore, there is a risk that knowledge workers might over-rely on these tools without adequately understanding their limitations. Critical thinking and human judgment remain essential in the interpretation and application of AI-generated insights.
Data Privacy Concerns: Users must be cautious when dealing with sensitive data as the tool could potentially be a point of data leakage. It's crucial that knowledge workers understand these risks and follow best practices for data privacy and protection.
Python-specific: Code Interpreter is specifically designed for Python. Users familiar with other programming languages might not find it as useful.
Limited Environment: Code Interpreter operates in a sandboxed environment, which has some restrictions. It cannot make API calls or access the internet.

Overall, Code Interpreter presents a significant opportunity for knowledge workers to augment their skills and productivity, but it also comes with challenges that need to be managed carefully.

Let’s Get Started! How to Access Code Interpreter?

Go to https://chat.openai.com/ and sign up via SSO of existing accounts such as Gmail or create a new OpenAI account
Once signed up, log in to ChatGPT and click on the three dots at the bottom right next to your profile picture and name to access “My plan”

Start Your First Code Interpreter Experiment: “Data Analyst”

Again, click the three dots right next to your profile at the bottom to access “Settings”
Click on “Beta features” to enable Code Interpreter via the toggle

Once activated, click on “Data controls” to turn off “Chat history & training”, just in case you plan to work with sensitive information that should rather not be used for training purposes by OpenAI

Get back to the chat interface and click on “GPT-4” at the top to activate Code Interpreter (beta)

Once activated, a “+” sign will appear left of the text box within the chat interface. You can now upload a variety of up to 100MB documents and add Python and/or human language instructions on what you want Code Interpreter to do with the data, see a simple example of my recent “10 Charts About the State of AI” dataset below

Subsequently, Code Interpreter transforms your intent into a custom Python script and runs it real-time to perform a task within the ChatGPT interface. In the “AI Startups & Investors” xls example above, Code Interpreter lists all tabs and summarizes the data contained therein

Next, it creates a variety of charts (bar charts, pie charts, scatter plots, etc.) together with some brief descriptions

At this point, you should be all set to start your own experiments with Code Interpreter. Similar to ChatGPT, I expect a range of best practices and guides to evolve in the near future. Excited to hear about your endeavors!

Share this article with others who might benefit.

Concluding thoughts

Code Interpreter stands as a compelling demonstration of AI's strength and its capacity to transform our interaction with data. Keep in mind that the future isn't merely an occurrence that we passively experience; it's a reality that we actively shape. Hence, play the game on the field, and let's welcome these emerging tools with open arms to construct a more efficient, effective, and inclusive future.

Stay driven,
Andre

PS: Comprehensive GPT-4 details got leaked here

PPS: I had the pleasure to join the

Reading Unicorns

podcast here and the

Starwatcher observatory

podcast here to discuss different angles of my path into VC, my belief in an “Augmented VC” approach, how VCs can start to become more data-driven, and a lot more

Thank you for reading. If you liked it, share it with your friends, colleagues, and everyone interested in data-driven innovation. Subscribe below and follow me on LinkedIn or Twitter to never miss data-driven VC updates again.

What do you think about my weekly Newsletter? Love it | It's great | Good | Okay-ish | Stop it

If you have any suggestions, want me to feature an article, research, your tech stack or list a job, hit me up! I would love to include it in my next edition😎