11 Next-generation AI

 

This chapter covers

  • Recommendations for preparing data for an AI model
  • Recommendations for which techniques to use
  • Properties the next-generation AI systems should have
  • Thoughts about what future AI systems should support

Building the AI solutions of the future requires us to address the current limitations in today’s systems. A key objective of this book is to provide a clear and honest assessment of the current state of AI because it’s only by understanding where we are today that we can chart a realistic path to the future. While media portrayals of AI often lean toward the sensational, my aim is to provide a balanced perspective. Much of the technology we find exciting and innovative today has actually been in development for over half a century. Although challenges remain, such as efficiency, cost-effectiveness, and adaptability, they present opportunities for growth and improvement as we continue on this exciting AI journey.

11.1 Data flexibility

11.2 Sampling

11.3 Elimination of irrelevant attributes

11.4 Data coherence

11.5 Lack of bias in data and algorithms

11.6 Feature engineering

11.7 Technique combination

11.8 Unsupervised learning

11.9 AI factory

11.10 Quality Assurance

11.11 Prediction reliability

11.12 Effective data storage and processing

11.13 Deployability and interoperability

11.14 Scalability

11.15 Resilience and robustness

11.16 Security

11.17 Explicability

11.18 Traceability and monitoring

11.19 Privacy

11.20 Temporal reasoning

11.21 Contextual reasoning

11.22 Causality inference

11.23 Analogical reasoning and transferability