Skip to main content

What AI Can and Cannot Do

Amazing Abilities​

Learning and Adapting​

AI systems can surprise us by:
  • Learning from just a few examples
  • Solving new types of problems
  • Understanding complex instructions
  • Improving through practice

Example: Like how ChatGPT can learn to write in different styles just by asking it to.

Memory and Understanding​

These systems can:
  • Remember conversations
  • Keep track of context
  • Handle long discussions
  • Connect different ideas

See how this works technically

Current Limitations​

Memory Constraints​

AI still has limits:
  • Can only remember so much at once
  • Might forget earlier parts of long conversations
  • Needs information repeated sometimes
  • Can get confused with too much information

Example: Like how we might forget details from the beginning of a long book.

Common Mistakes​

Making Things Up​

Sometimes AI can:
  • Create false information
  • Mix up different facts
  • Sound confident about wrong answers
  • Misunderstand complex questions

Learn about preventing mistakes

Biases and Limitations​

The system might:
  • Reflect biases from training data
  • Struggle with current events
  • Miss cultural context
  • Misunderstand nuanced topics

Measuring Success​

How We Test AI​

We check performance through:
  • Accuracy testing
  • Human evaluation
  • Consistency checking
  • Real-world testing

See how testing works

Looking Forward​

Future Improvements​

We're working on:
  • Better memory systems
  • More accurate responses
  • Reduced biases
  • Broader understanding

Explore future developments

Want to learn more?