AI Job Interviews - another good example of bias in ML 🤦‍♂️

Two journalists tested some AI tools for assessing job candidates. Even when the candidate read a Wiki article in German instead of answering questions in English, the AI systems gave them good scores 🤷‍♂️

Let's unpack 👇

The Setup 🔬

The journalists created a fake job posting on two AI interview platforms. They specified the traits of the ideal candidate and provided the questions that need to be answered during the interview.

Then they started experimenting... 👇

The Positive Test ✅

One of them did a fake interview giving all the right answers and predictably got very high scores - 8.5 out of 9 👍

Then she tried something different... 👇

The Negative Test ❌

In a second interview, instead of answering the questions in English, she just read the article on psychometrics from the German Wikipedia 😁

One system gave her a score of 6 out of 9, while the other determined she is a 73% match for the job.

Oops... 👇

What happened? 🔍

Interestingly, one of the systems generated a transcript which was obviously meaningless.

This means that the machine learning model behind the tool likely captured nuances of the intonation of the speaker instead of the meaning of the actual words.

👇

Do machines need emotions to be “really intelligent”? Our best reference for an AGI (human brains) points in that direction.

If so, how on earth would one synthesize or encode “sadness” or “joy” for starters? 🤖?❤️

Join me and see how far we get! 🧵

Thx @AlejandroPiad, @svpino

I’ve worked on the topic of emotions + AI for a while now with the goal to “give the Tin Man” a heart.

I haven’t... yet :)

But I did learn a lot of “emotional engineering” along the way that you may find interesting or inspire you new RL approaches.

In this thread I’ll cover many elusive psycho-bio-neuroscientific concepts from a computational perspective. (Bear in mind I’m a computer scientist, so feedback from people in those fields is welcome!)

Ok, so firstly, HOW are emotions useful for us living things?...

Our behaviours are (very strongly) driven by emotional forces, often prevailing over our more "rational" processes.

They also spark and feed our learning, focusing it on our most intense emotional episodes ("that hurt, I'll avoid it; that was great, I want more").

Furthermore, when expressed externally, emotions also provide a primordial language that allows inter- and intraspecies communication. (Isn’t it fantastic how watching a nature documentary we can tell how animals feel from sounds and body language?)

All the @madewithml machine learning fundamentals & MLOps lessons are released!

- 🛠 Project-based
- 💻 Intuition & application (code)
- 🏆 26K+ GitHub ⭐️
- ❤️ 30K+ community
- ✅ 47 lessons, 100% open-source

https://t.co/XIhD3wl1DA

🧵 Thread on details & lesson highlights 👇

Who is this course for?

- 💻 Software engineers / Data scientists looking to learn how to responsibly create ML systems.
- 🎓 College grads looking to learn the practical skills they'll need for the industry.
- 🚀 Product Managers who want to develop a technical foundation.

We start with lessons on the fundamentals of ML through intuitive explanations, clean code and visualizations.

📚 Foundations
- Python (variables, functions, classes, decorators)
- NumPy (numerical analysis)
- Pandas (data analysis)
- PyTorch (operations, gradients)

Then we dive into implementing basic ML algorithms 1⃣ from scratch then 2⃣ in PyTorch. Starting from simple models → complex models.

📈 Modeling
- Linear Regression
- Logistic Regression
- Neural Networks
- Data Quality (⚠️ very important)
- Utilities (for loading and training)

We wrap up the fundamentals by implementing deep learning algorithms in PyTorch.

🤖 Deep Learning
- CNNs
- Embeddings
- RNNs
- Transformers

💡 We motivate the need for specific architectures and additional complexity as we implement each method.

How do you prioritize a roadmap?

First, take a step back and remember that a business is just a group of people solving problems for other people in exchange for money. Your job is to (1) figure out the problems, and (2) solve them in the most efficient way possible. Thread 👇

1/ Step 1: Figure out the problems -- get the ground truth:
* Talk to customers
* Talk to employees who talk to customers (CX, sales)
* Observe your customer's behavior, through data and user research
* Use the product yourself
* Look into what caused users to churn

2/ Step 2: Solve problems in the most efficient way possible -- least work for most impact:
* Develop hypotheses around what you believe will solve the problems
* Ideate solutions to these hypotheses (see link)
* Put these ideas into a spreadsheet

3/ Rate each idea from 1-5 across two attributes: (1) Impact: How impactful it'll be in solving the problem, and (2) Effort: How much work it'll take to execute. Guesstimate -- what matters most is relative differences.
* Sort the spreadsheet by ratio of Impact / Effort, discuss

4/ How do you estimate Impact?
Every decision is somewhere along this spectrum:
Gut <--------------------> Complete data
Knowing this, use all of the data at your disposal (quant, qual), plus your team's experience, to inform the estimate. You are smart people, trust yourselves.

Beginners guide to start with machine learning.

What you need:

1. A programming language
2. A place to write and run code
3. A way to deal with data
4. A way to visualize results
5. A cool algorithm
6. An introductory course
7. A project to solve

Here are my recommendations.

Most people recommend Python to start.

I do too.

You can do machine learning with many different languages, but today, Python is the best option.

My advice: Get comfortable writing code before looking into machine learning.

You don't have to be a great developer to start.

But if you are, it helps.

Everything you know about building software is helpful to build machine learning systems.

Good developers have the odds in their favor.

"How much Python do I need to know before starting?"

If you ask, you aren't ready yet.

Ideal scenario: You should be comfortable building software. Most people get here within a year.

Let's move on to the 2nd item from the list: A place to write and run your code.

Get familiar with Google Colab.

• It's free.
• Requires no setup to start.
• It's available from anywhere.

Alternative: Kaggle.

0/ Nothing pisses me off more than Lawyers ripping Founders off when putting investment docs together.

The worst part is most Investors aren’t helpful - 85% push the bill to Founders.

As an ex-lawyer, I saw all the inside tricks.

Here's how to reduce your legal bill by 90%:

1/ First, it’s important to understand how lawyers make $.

A legal bill has nothing to do with the end deliverable.

Wait what? That’s right. Lawyers make money via the billable hour.

Regardless of quality, you get charged based on how many hours the lawyer(s) spent with you

2/ The second cost variable is hourly rate.

Hourly Rate is a function of seniority of the lawyer you are working with.

Why is this important? Because the hourly rates go up FAST at top law firms.

Junior Associates = ~$400/hr
Senior Associates = ~$1000/hr
Partners = $1000+

3/ Most financings are very routine.

This was obvious at Seed (hence YC Safe), but it's also the case at A.

The problem is, as a Founder:

(a) You don't know what you don't know
(b) You don't want to mess it up
(c) You want to close fast
(d) You want to move on to your biz

4/ And that’s how you get trapped.

If you let the lawyers run it, a financing process will easily cost you $150k+.

You need to do 2 things: (1) pick the Lawyer and (2) establish the ground rules.

K, I got @arvidkahl 's book The Embedded Entrepreneur.

I'll be live-tweeting quotes as I read it over the next couple of days.

Here we go.

"At first, communities are hunting grounds for great opportunities, then they turn into places to build meaningful relationships, and later they'll be the fertile soil in which you will grow your business." - @arvidkahl, The Embedded Entrepreneur

"Every day, a developer some where considers building another variation of an existing tool. They build this because they need it themselves,
but they don't consider if their need is a commonly shared one." - @arvidkahl, The Embedded Entrepreneur

"No business was ever successfully built by a founder who didn't care about the people they were selling their product to." - @arvidkahl, The Embedded Entrepreneur

"For bootstrappers, a market has to be both large enough to sustain your business and small enough not to attract giant competitors." - @arvidkahl, The Embedded Entrepreneur

Machine learning education is broken.

If you are preparing for a research position, you are good. If you are looking to get out there and start solving problems, not even close.

Here are some thoughts so you can get ahead.

↓

Most classes, courses, and books cover the same road.

They start with a dataset. They finish with a working model. The focus is always on everything that happens in between.

Dataset → Model.

This is great, but not enough.

Real-life situations rarely start with a dataset, and they never end after you finish building your model.

Applying machine learning successfully is hard.

Here are a few examples that you should keep in mind.

First challenge: Properly framing up the problem.

If you don't understand the problem, you can't determine what data you need. If you don't understand the data, you can't build a good model.

I've never seen a company that had their data ready to go.

In fact, most of them don't even have data at all and need you to determine what exactly they should start collecting.

You usually have to go Problem → Potential Solution → Data.