My Second Python Program

By Wayne Brown

Learning Python


In this blog, I try to provide examples of my actual experience with AI, warts and all. You may have heard that programming is obsolete. Lets see why this might be true.

Following a long tradition, my first program simply said, "Hello World." Without going into details, this took me days to get working In spite of the fact that I have written computer code for money for 50 years, I suck at it.

For my second effort, I had AI do the coding. What I asked for was this:

Write Python code to import history.csv into a frame. Name columns "title" and "date". Copy the title column to a list called raw_titles. Make a new list strip_titles from raw_titles, where the text contains only raw_title text up to the first ":" Save unique elements of strip_titles in alphabetical order as a new file csv called unique_titles.

 That request reflects some vague idea of what's possible and what I want, but it's English, not Python. Perplexity gave me this:

import pandas as pd

# Load the CSV file into a DataFrame and specify the column names

df = pd.read_csv('history.csv', names=['title', 'date'], header=0)

# Copy the 'title' column to a list called raw_titles

raw_titles = df['title'].tolist()

# Create a new list strip_titles from raw_titles, where the text contains only raw_title text up to the first ":"

strip_titles = [title.split(':')[0] for title in raw_titles]

# Get unique elements from strip_titles and sort them alphabetically

unique_strip_titles = sorted(list(set(strip_titles)))

# Save unique elements of strip_titles as a CSV file called unique_titles.csv

unique_titles_df = pd.DataFrame(unique_strip_titles, columns=['title'])

unique_titles_df.to_csv('unique_titles.csv', index=False)

Note that Perplexity helpfully provides comments, which educate me about how Python solves the problem. In fact, AI can give me a very detailed explanation of how all this works.

But why do this?

You can get a list of movies you have seen on Netflix (left as homework for the reader) but it has entries for each episode of long-running series like "Broken Bad." That's why I need to get rid of the ":" and ask for unique titles. 

But why do this?

The resulting file (unique_titles.csv) can be directly dragged into a Perplexity prompt to become the "context" for a discussion. See the result here. Coding 7 lines of Python to get this result did involve a bit of a learning curve, but no actual coding was required. This is truly my 2nd Python program, and I really didn't write it. I'm still personally back at "Hello World." The key insight was actually the need to hand Perplexity as small a file as possible. There is a limit to how much "context" an AI can "remember". Today. There is a trend to increase this limit from a few hundred characters to megabytes.

For your information, this is what "unique_titles" looks like:

13 Going on 30

1899

2012

3

3%

365 Days

A Beautiful Life

A Man of Action

A Perfect Murder

A Quiet Place Part II

A Trip to Infinity

After Life

After We Fell

Age of Tanks

Agent Elvis

Alexander

Alias Grace

All Quiet on the Western Front

All the Light We Cannot See

All the Money in the World

Always Shine

American Gangster

American Girl

American Woman

Anatomy of a Scandal

Ancient Apocalypse

And so on ...

Netflix provided me with a file that looks like this:

 "Griselda: Limited Series: Adios, Miami",2024-02-18

Griselda: Limited Series: Paradise Lost,2024-02-17

Griselda: Limited Series: Middle Management,2024-02-16

Griselda: Limited Series: Mutiny,2024-02-16

Griselda: Limited Series: Lady Comes to Town,2024-02-16

Raël: The Alien Prophet: Limited Series: Raël Superstar,2024-02-16

Raël: The Alien Prophet: Limited Series: Genesis,2024-02-16

Working: What We Do All Day: Limited Series: The Middle,2024-02-16

ROMA,2024-02-16

Griselda: Limited Series: Rich White People,2024-02-15

American Girl,2024-02-14

No Limit,2024-02-14

365 Days,2024-02-14

Once Upon a Time in Hollywood,2024-02-14

Always Shine,2024-02-13

The Deepest Breath,2024-02-13

Inside the Mind of a Cat,2024-02-12

Talk to Me,2024-02-12

Lead Me Home,2024-02-12

War Dogs,2024-02-11

Outer Banks: Season 1: Pilot,2024-02-09

Safe: Season 1: Episode 8,2024-02-09

Safe: Season 1: Episode 7,2024-02-09

Safe: Season 1: Episode 6,2024-02-08

Safe: Season 1: Episode 5,2024-02-08

Safe: Season 1: Episode 4,2024-02-08

Safe: Season 1: Episode 3,2024-02-07

Safe: Season 1: Episode 2,2024-02-07

Safe: Season 1: Episode 1,2024-02-07

===============

This is not foolproof.  Perplexity can't find titles that are obviously in the file for totally opaque reasons but then changes its "mind", apologizes and makes up stupid excuses. It's very evident that AI's logic is not necessarily what a human would follow. There is no way to know that until things go sideways. It's quite possible that errors will go undetected. Also, slight variations in the approach, such as reading from an Excel file rather than CSV, land you in "import Hell," where the required module is/is not installed. This had me stuck or many days before I even attempted the seemingly magical solution here. Getting your modules right is the next step after "Hello World".

The last time I was paid to do anything with computers, my group was called "the plumbers." Our job was to tie big applications together so they would "talk" to each other. For example, an application that accessed the library would need to talk to the application that logged people in and the application that said whether access was permitted. There was little or no "programming" involved. Talking to AI is a bit like that. It was the most boring job I had in over 50 years. It was a relief when the crash of 2008 caused half of the department to be laid off.

All this can be funny until you realize that AI is being trusted in some applications with life-and-death situations. For example, Tesla cars running on "fully automatic" happily stop for children and then run them over. 

You can also see this in the ordinary jobs I give AI every day. More on that later ...

 

Comments

Post a Comment

Popular posts from this blog

Dennis Hoffman and The Nature of Reality

By Wayne Brown
Image
Perplexity Massive data loss.... The great math mystery "Math works so well to describe reality because math is all there is." just 32 numbers and a bit of math Lex Fridman & Dennis Hoffman The Case Against Reality (In Kindle) This may be the wrong question: Does math work so well because it's how we perceive reality or is math a fundamental property of the Universe? Missing is the idea that consciousness itself is essential and that a mathematical description of consciousness could be "fundamental" - more "fundamental" than what we think of as "reality." "The elegance of math meets the messiness of reality" - but this is a clue to how we simplify reality for efficiency. 
Read more

Panic Part 6 - The IPCC Summary for Policymakers

By Wayne Brown
Image
Love it or hate it, the discussion revolves around this report Bottom line: I accept the IPCC report as a kind of "hub" around which my view of climate issues must turn. It is important to understand what it is saying, along with its limitations. Legions of cherry pickers have descended on this report - many to raise alarm, many to rebut, many to blow smoke over it. It is not difficult for a reasonably intelligent person to read the report itself, especially to examine the baskets of cherries emerging from the forest. Generally speaking, I have found critics of the report to be highly misleading. To call the report "alarmist" is perhaps simply to admit that there is no shortage of worrying facts in the report. For example, the frequently cited concern of the "alarmists" - namely sea-level rise - is completely well documented. To me, the "let them drown" response at COP26 forms the basis of my assumption that the politicians of the world will not...
Read more

A Challenge to the "Settled Science" Meme

By Wayne Brown
Image
I've read at least a dozen books on the climate issue in the last month. This is the best. Koonin is not just another "skeptic." His approach to the matter rises above "controversy" and depends on evidence. Yet, he acknowledges where the panic is coming from. A lecture by Koonin covers much of the key material and provides color versions of the slides, which are black and white in the book. Another interview, without the slides, is here . In reading IPCC reports, it's evident that the "summaries" don't "summarize" the actual science. The ultimate summary, the speech by the head of the IPCC , bristles with inaccuracies and has no message beyond panic. As a senior climate scientist, Koonin provides some startling stories of the "blow-back" he gets by simply asking for an independent review of IPCC conclusions. In fact, there was a brief attempt to cripple such a review by denying federal funds to any study that challenged t...
Read more