OpenAI: Code Interpreter here
...step forward in ‘software eating AI’
This week has OpenAI making available its highly anticipated Code Interpreter Plug-in software for subscribers of ChatGPT Plus, using its GPT4 LLM AI. Here’s why it’s an important step in what we’ve discussed before in the process of ‘AI eating Software’.
“It lets ChatGPT run code, optionally with access to files you've uploaded. You can ask ChatGPT to analyze data, create charts, edit files, perform math, etc.”
“We’ll be making these features accessible to Plus users on the web via the beta panel in your settings over the course of the next week.”
The enthusiasm and anticipation for this capability has been wide and deep, and users of all stripes have had positive reactions to what’s possible going forward. As one developer explains:
“It makes the AI much more versatile. A remarkable number of problems can be solved with code, and GPT-4 is very good at figuring out when to use Code Interpreter in novel and interesting ways.”
And a practical review of what’s possible with it:
“This is just scratching the surface of Code Interpreter, which I think is the strongest case yet for a future where AI is a valuable companion for sophisticated knowledge work.”
“Things that took me weeks to master in my PhD were completed in seconds by the AI, and there were generally fewer errors than I would expect from a human analyst. Human supervision is still vital, but I would not do a data project without Code Interpreter at this point.”
Another reviewer adds:
“One of the most intriguing applications of Code Interpreter is in data science, where it has been described as operating at an “advanced level.” It can automate complex quantitative analyses, merge and clean data and even reason about data in a human-like manner. “
“The AI can produce visualizations and dashboards, which users can then refine and customize simply by conversing with the AI. Its ability to create downloadable outputs adds another layer of usability to Code Interpreter.”
And this reviewer on combining disparate code libraries in truly versatile ways:
“Beside generating code, Code Interpreter (CI) can analyze the output and use it in another function. This means that you can string together different sections of code, taking the output of one and feeding it to another. The Pac-Man gif above was made by having CI use an algorithm to generate a maze, convert the maze into blocks, use an algorithm to find the exit, make it look like Pac-Man and then generate a gif.
Previously when I used ChatGPT to create code this involved taking the output and putting it into another environment to test it. Now you can do a lot of development inside ChatGPT without leaving the UI. When you add that to its ability to analyze data, such as spreadsheets, and then doing functions like graphing, you have a very powerful tool for both coding and research.”
As I outlined in the piece ‘AI eats Software’ a few weeks ago, the practical opportunities around AI are when users are able to blend the deterministic characteristics of traditional software, with the probabilistic capabilities of LLM AI.
“And that is, to in real time, take our questions, then run unimaginably complex loops of statistical analysis with them, against every imaginable set of digital data captured thus far, and provide predictive “answers” that feel like the “AI” knows us and understands us.”
The current reviewers of Code Interpreter Plugin are already pointing to these ‘AI Moments’ made possible by this prosaic next step in running Python code and data files in a sandboxed ChatGPT prompt query session. We are at the beginning of what’s possible as these capabilities are developed to scale.
OpenAI Founder/CEO Sam Altman had said a few weeks ago that ChatGPT Plug-ins had yet to find their ‘product-market-fit’. Code Interpreter is an important step towards getting there. It’s still very early days. Stay tuned.