Abstract

This paper studies the limitations and socio-political implications of data-centric AI development process. It argues that the contemporary AI development is limited by its own definition of the field whereby it has become oblivious to several human realities. The paper argues that the big digital-eyes project marks the beginning of a new class of machines that can be called ‘Affective-Machines’. In conclusion, the paper argues that there is a need to shift towards a trans-disciplinary systems approach in order to integrate perspectives from various disciplines while trying to negotiate the socio-political issues arising.

Keywords: 1) Autonomous vehicles; 2) Trans-disciplinary approach; 3) Data-Centric AI; 4) Memory-Machines; 5) Affective-Machines

Table of Contents

1.     Introduction

2.     Limits of Contemporary AI Development

2.1.  Explainability

2.2.  Reproducibility

2.3.  Adaptability

2.4.  Monitoring and Modulation

3.     Why Big Digital-Eyes on Autonomous Vehicles?

4.     Socio-Political Implications of the Contemporary AI systems

5.     Need to Shift Towards Human-AI Hybrid Systems Approach

6.     Conclusion

7.     References