Dan's Brain

On AI Education

Notes

What just happened in AI and how it is being misunderstood

  • two main ways to tackle a problem
    • model-based
      • represent and reason
      • using logic and probability
    • function-based
      • fit function to data
        • just give me data, make a blackbox
        • no need to understand exactly what’s going on under the hood
      • using Neural Network
      • taught in the classrooms
      • good for tasks usually associated with perception or cognition
  • eras of AI
    • knowldge representation and reasoning
      • very CS based
    • machine learning
      • very Statistics and Optimization based
    • neural networks
      • mostly Engineering based
  • as we are moving forward the background of researchers are being diluted
  • now the tech is being deployed and it’s hitting walls
    • there is need of logic skills in the next generation of AI research
    • solid computer science education is needed