AI Players Learn Bad Habits from Watching Failures, Study Finds
AI that mimics failure could make games harder and training less effective.
Deep Dive
arXivLabs is a framework that lets collaborators develop and share new arXiv features directly on the website. Both the individuals and organizations working with arXivLabs have embraced and accepted arXiv's values of openness, community, excellence, and user data privacy — and arXiv is committed to those values, working only with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? The article points readers to learn more about arXivLabs.
Key Points
- AI players often repeat moves they've just seen fail, unlike humans who learn from mistakes.
- This happens because AI is trained on examples that include failures, and it copies them without understanding why they failed.
- The problem could make games less enjoyable and AI training less efficient, wasting time and money.
Why It Matters
If AI can't learn from failure, it may repeat mistakes in games, work, and daily life, costing time and money.