CS2125 Paper Review Form - Winter 2019

Reviewer: Mohammad Rashidujjaman Rifat

Paper Title: Machine learning, social learning and the governance of self-driving cars.

Author(s): Jack Stilgoe

1) Is the paper technically correct?
 [✓] Yes
 [ ] Mostly (minor flaws, but mostly solid)
 [ ] No

2) Originality
 [✓] Very good (very novel, trailblazing work)
 [ ] Good 
 [ ] Marginal (very incremental)
 [ ] Poor (little or nothing that is new)

3) Technical Depth
 [ ] Very good (comparable to best conference papers)
 [✓] Good (comparable to typical conference papers)
 [ ] Marginal depth
 [ ] Little or no depth

4) Impact/Significance
 [✓] Very significant
 [ ] Significant
 [ ] Marginal significance.
 [ ] Little or no significance.

5) Presentation
 [✓] Very well written
 [ ] Generally well written
 [ ] Readable
 [ ] Needs considerable work
 [ ] Unacceptably bad

6) Overall Rating
 [✓] Strong accept (award quality)
 [ ] Accept (high quality - would argue for acceptance)
 [ ] Weak Accept (borderline, but lean towards acceptance)
 [ ] Weak Reject (not sure why this paper was published)


7) Summary of the paper's main contribution and rationale
   for your recommendation. (1-2 paragraphs)

This paper argues that the use of machine learning in developing self-driving cars pose an essential governance challenge. In particular, this paper scrutinizes in details the emerging politics of machine learning and problems of algorithmic processes and outcomes. Finally, this paper recommends governance options that seek to prioritize social learning, focusing in particular on the sharing of data.

The paper starts by making the case that the self-driving car will be in the state of beta option for a long time. The reason for this is it cannot absolutely envision social rules. Consequently, when an accident occurs, it is painted as an unintended consequence and the human becomes the moral crumple zone. The paper develops its argument about the problematic issues of machine learning to implement self-driving car systems. Machine Learning systems are designed as black-boxed; nobody except for companies can understand the ML algorithms used to develop a system. Besides, machine learning systems are yet not capable of handling many ethical situations. The paper argues for transparency in ML algorithms. The paper further makes its arguments stronger by explaining the case of Tesla.  

In recommending to improve self-driving car development, the paper suggests taking the role of ‘social’ seriously. To do so, autonomous vehicle research can draw on educational psychology, emphasizing on longitudinal learning. The other recommendations include broadening the view of social, improving the interpretability of algorithm, improving the interpretability of algorithm, public participation not as education, but as democracy, public participation not as education, but as democracy.


8) List 1-3 strengths of the paper.  (1-2 sentences each,
identified as S1, S2, S3.)

S1. The paper is a strong critic of machine learning in the case of using it for implementing self driving car. It will help to improve the autonomous vehicle research by taking social issues seriously.
S2. The paper refers to a substantial and strong set of articles that can benefit machine learning systems in general.
s3. This is a very well written paper with essential connections to necessary literature.

9) List 1-3 weaknesses of the paper (1-2 sentences each,
identified as W1, W2, W3.)

W1. The paper lacks in providing pragmatic suggestions in several places. Though, it is not bound to give any suggestion.
W2. The paper could have made direct critic of using ML in a particular test case for self driving cars.