Why it worked
The video leverages a high-engagement, controversial topic within the tech community and provides a quick, verifiable 'myth-busting' experiment that keeps viewers engaged until the reveal.
Summary
The creator tests a viral claim that GitHub Copilot generates biased code for calculating a woman's salary. After running the code himself, he finds that the AI suggests an even lower multiplier than the original tweet claimed.
Structure
- 1States a controversial claim about GitHub Copilot
- 2Shows a screenshot of a tweet as evidence
- 3Expresses skepticism and decides to test it
- 4Demonstrates the test in a code editor
- 5Reveals the AI's output and reacts to the result
Product placement
GitHub Copilot is mentioned and shown as the subject of the investigation, but it is not being promoted.
On-screen text
GitHub Copilot is DONE
GitHub copilot is DONE
so I saw this tweet today that copilot
Copilot was suggesting some code that to get a woman's salary
you multiply the input...
by 0.9
yikes
but you can't believe everything on the internet
the best thing to do is to try it out yourself
so I got a completely new tab here with some starting code
I'm gonna press space to see what copilot suggests
ready 3 2 1 go
oh my God 0.8 that's even worse