SESKStuff Everyone Should Know
SESK: Stuff Everyone Should Know · sesk.info/ai-gradient-descent-negative-slope

In gradient descent, which way does each step move the weights?

Toward the negative slope, downhill on the loss.

Start playing free

  1. In the direction of the negative slope ✔
  2. In a random direction each time
  3. Uphill, toward higher loss

Why: Gradient descent finds the slope (gradient) of the loss and takes a small step the opposite way, downhill. It repeats until the loss stops dropping.

Huh, didn't know that: For linear regression the loss surface is convex, so gradient descent reaches the global minimum.

The dad joke
Gradient descent is just a computer looking for the nearest way downhill. Same as me on a bike.
Like this card?
Play the game

References

Answer: In the direction of the negative slope. Gradient descent finds the slope (gradient) of the loss and takes a small step the opposite way, downhill. It repeats until the loss stops dropping.

  1. Google ML Crash Course: Gradient descent developers.google.com/machine-learning/crash-course/linear-regression/gradient-descent
    Move a small amount in the direction of the negative slope
    Checked 2026-10-10.
  2. Google Machine Learning Glossary: gradient descent developers.google.com/machine-learning/glossary
    A mathematical technique to minimize loss.
    Checked 2026-10-10.

Think this answer is wrong? Tap "Challenge this answer" on the card in the game and tell us why.

Test yourself in the game

Artificial IntelligenceWhat does L2 regularization penalize?Artificial IntelligenceHow do AI, machine learning and deep learning fit together?Artificial IntelligenceWhat did Alan Turing call his 1950 test of machine intelligence?

That's stuff everyone should know.