12 New Code Interpreter Uses (Image to 3D, Book Scans, Multiple Datasets, Error Analysis ... )

Use Cases for Code Interpreter

In this video, the speaker discusses various use cases for their code interpreter. They demonstrate how it can be used to create a 3D surface plot, analyze text sentiment and tone, write an essay, and find correlations between data sets.

Creating a 3D Surface Plot

• The code interpreter can create a 3D surface plot by analyzing the RGB of pixels in an image.

• It can also reflect objects in the image such as buildings.

• The interface is user-friendly and accessible through chat GPT.

Analyzing Text Sentiment and Tone

• The speaker uploaded Anna Karenina into the code interpreter to analyze all mentions of England and discover how it was perceived in the book.

• The model found relevant quotes and analyzed their tone and sentiment.

• It then generated a two-thousand-word reflection on the presentation of England in Anna Karenina.

• This demonstrates that the model can analyze large amounts of text to generate insights.

Writing an Essay

• Using The Death of Ivan Ilyich by Leo Tolstoy as input, the code interpreter wrote an epilogue to the story.

• It also outputted this essay as a PDF file.

Finding Correlations Between Data Sets

• The speaker uploaded four data sets into the code interpreter to find any correlations between them.

• There doesn't seem to be a limit on how many files you can upload at once.

• The model asked for guidance on what kind of analysis or specific questions were needed before proceeding with its analysis.

Correlations and ASCII Art

In this section, the speaker discusses how they used GPT-3 to find correlations in data sets and create ASCII art.

Finding Correlations

• GPT-3 can find correlations in data sets.

• The speaker gave an example of a moderate positive correlation between income inequality and murder rates.

• GPT-3 can provide plausible explanations for correlations found in data sets.

Creating ASCII Art

• The speaker received requests to create ASCII art using GPT-3.

• They provided an example of creating ASCII art from an image using GPT-3.

Analyzing Data Sets

In this section, the speaker demonstrates how they used GPT-3 to analyze large data sets for anomalies.

Analyzing CSV Files

• The speaker tested if GPT-3 could spot errors in a mass passive CSV or Excel file.

• They uploaded a file with thousands of data points and asked GPT-3 to find any anomalies by looking for implausible percent changes year to year.

Writing Poetry with AI

In this section, the speaker shows how they used GPT-3 to write poetry about dystopian landscapes.

Writing Poetry

-The speaker asked GPT - 3 to write a sonnet about a more full AI reflecting on a dystopian landscape.

• GPT-3 used an Edge detector to overlay the poem in the background of an image.

• The final result was a sonnet that reflected on the loss of humanity and the sorrow of a machine left behind.

Analyzing Poll Data

In this section, the speaker demonstrates how they used GPT-3 to analyze poll data.

Analyzing Poll Data

• The speaker pasted in poll data from a website with confusing formatting.

• They asked GPT-3 to extract data for the 2024 Republican Presidential nomination and analyze trends in the data.

Using GPT-4 Code Interpreter for Political Prognostication

The speaker discusses how they used a messy data set to do political prognostication using GPT-4 code interpreter. They also discuss the accuracy of the results and provide a link to the data set for further checks.

Analyzing Data and Providing Insights

• Used an unclean data set to analyze other races and pasted in an article from Politico.

• GPT-4 analyzed the article and polls, providing smart and nuanced insights.

• Uploaded fictional food company data set based in Boston and New York, asking for six actionable insights.

• GPT-4 provided six visualizations along with insights and actions that could be taken as a CEO.

Plausible Explanations for Findings

• Asked to offer plausible explanations for each finding based on personal knowledge of the world.

• GPT-4 offered explanations such as higher sales in east region due to higher population density or better distribution networks.

Suggesting Other Company Data Sets

• Asked to suggest six other company data sets that would be helpful to access to test suppositions.

• GPT-4 suggested files that would help with analytics, but privacy concerns should be considered before uploading them.

Outputting PowerPoint Slides Directly with Code Interpreter

The speaker discusses how code interpreter can output PowerPoint slides directly, which is cool even though it may become redundant when Microsoft 365 copilot rolls out.

Outputting Visualizations into PowerPoint Slides

• Code interpreter can output visualizations and analysis directly into PowerPoint slides.

Comparing Code Interpreter and Wolfram on Differential Equations

The speaker compares code interpreter and Wolfram on differential equations, finding that code interpreter has some serious oomph behind it.

Testing Code Interpreter and Wolfram on Differential Equations

• Tested both code interpreter and Wolfram on differential equations.

• Both got it right, but Wolfram required a paid option for step-by-step solutions.

Differences Between Code Interpreter and Wolfram

• Found differences between the two when testing them with a challenging mathematics question.

• Code interpreter consistently got it right while Wolfram did not.

Uploading Tolstoy's "The Death of Ivan Ilyich" to Test GPT-4

The speaker discusses how they uploaded Tolstoy's "The Death of Ivan Ilyich" to test GPT-4, changing one phrase in one line out of about 23,000 words.

Testing GPT-4 with Tolstoy's "The Death of Ivan Ilyich"

• Uploaded the entire short story by Tolstoy and changed one phrase in one line.

• Asked if anything about the daughter in the story seemed strange.

• GPT-4 eventually spotted the mistake despite being given little help.

Money and Smart Prompt Engineering

In this section, the speaker talks about the potential use cases of a new technology that is going to be rolled out soon. The speaker also mentions the advanced prompt engineering like smart gbt and tree of thought prompting.

Potential Use Cases

• The new technology is going to have a lot of use cases both professionally and personally.

• The speaker does not know when they are going to roll this out more widely.

Advanced Prompt Engineering

• Smart gbt and tree of thought prompting are examples of advanced prompt engineering that can be used with this technology.

• If you want to learn more about the code interpreter plugin, check out the other video by the speaker.