The Ethical AI Guy

Make AI ethics practical for builders and understandable for everyone.

Subscribe ↓

[The Ethical AI Guy @TheEthicalAIGuy]

Fairness in AI is not a vibe; it is a gate.

Here are 5 fairness checks I consider for any model:

1. Worst-group metric (Demography Parity/Disparate Impact/EOd)
2. Intersections (not just single attributes)
3. Min-N for power
4. 95% CI (bootstrap)
5. Pass rule + KPI guardrail

[DataCamp @DataCamp]

1/ Here’s how pandas groupby() works:

It groups rows in your DataFrame by a key (e.g. product, region, team) and lets you run calculations on each group. Think of it as SQL’s GROUP BY—but with more flexibility. 👇

[Samuel S. @itz_samietex]

I am glad to announce that after several months of exploring the often "vague" idea around ethical and responsible AI solutions, I have successfully completed the AI Ethics program with Turing College.

Cheers to responsible AI solutions🥂🎉.

[Samuel S. @itz_samietex]

It's so crazy to realize that I have written and published 18 articles on @DataCamp!🙂

I believe these articles will be useful for folks just starting their data analysis journey, so I will be sharing them on here.

A Thread🧵

#DataScience #Python #DataAnalysis

[The Ethical AI Guy @TheEthicalAIGuy]

I’ll break down each check with examples this week. Follow + turn on notifications.

[The Ethical AI Guy @TheEthicalAIGuy]

Fairness in AI is not a vibe; it is a gate.

Here are 5 fairness checks I consider for any model:

1. Worst-group metric (Demography Parity/Disparate Impact/EOd)
2. Intersections (not just single attributes)
3. Min-N for power
4. 95% CI (bootstrap)
5. Pass rule + KPI guardrail

[The Ethical AI Guy @TheEthicalAIGuy]

Follow for more contents on AI Ethics

#AIEthics #ResponsibleAI #MLOps

[The Ethical AI Guy @TheEthicalAIGuy]

Fairness in AI is not a vibe; it is a gate.

Here are 5 fairness checks I consider for any model:

1. Worst-group metric (Demography Parity/Disparate Impact/EOd)
2. Intersections (not just single attributes)
3. Min-N for power
4. 95% CI (bootstrap)
5. Pass rule + KPI guardrail

[Andrej Karpathy @karpathy]

I think congrats again to OpenAI for cooking with GPT-5 Pro. This is the third time I've struggled on something complex/gnarly for an hour on and off with CC, then 5 Pro goes off for 10 minutes and comes back with code that works out of the box. I had CC read the 5 Pro version and it wrote up 2 paragraphs admiring it (very wholesome). If you're not giving it your hardest problems you're probably missing out.