Abstract / Description of output
Anticipation occupies a special role in Artificial Intelligence (AI), not only because replicating human anticipatory processes is naturally a part of the aim to replicate human intelligence, but also because the design, implementation, and evaluation of AI systems involves a mix of several, interdependent anticipatory processes, some of which are carried out by human designers and users, and others by the AI system. This chapter provides an introduction to key AI technologies, investigates to what extent they involve anticipatory processes, and explores the consequences of such anticipation on the predictability of AI systems for humans. Our analysis suggests that the relationship between human and AI anticipation is complementary - the more pronounced the anticipatory capabilities of the AI system, the harder it may be for humans to anticipate their behaviour. If this turns out to be correct, building safe, responsible, and ethically sound AI will require developing more understandable and explainable AI methods in the future.
| Original language | English |
|---|---|
| Title of host publication | Handbook of Anticipation |
| Editors | R.~Poli |
| Publisher | Springer International Publishing |
| Pages | 1-33 |
| Number of pages | 27 |
| ISBN (Electronic) | 978-3-319-31737-3 |
| DOIs | |
| Publication status | Published - 10 Jul 2017 |
Keywords / Materials (for Non-textual outputs)
- Artificial Intelligence
- Anticipation
- Anticipatory process
- Computational modelling