The AI Accountability Challenge

Topic: If an AI makes a mistake, who should be responsible?

Introduction

There is no denying the fact that AI has not just revolutionized but redefined humanity’s capabilities. AI has become integral and its presence can no longer be overlooked. Companies across the fields of medicine, IT and even education are investing in AI. AI is the future and the future has truly arrived.

Artificial Intelligence might seem the best candidate to complete algorithmic programming or workflows. Like a human being in any rung of the corporate ladder, Artificial Intelligence can make mistakes. Mistakes are unavoidable even for modern-day technological marvels. But that is when we begin to tackle the elephant in the room: Who must take the blame if AI makes an error?

The Elephant in the Room: AI Dilemma

If humans make an error, they will be reprimanded and trained depending on the severity. But if Artificial Intelligence happens to make an error, then a written warning would do no good and neither would a salary cut. So, if Artificial Intelligence makes a mistake, who is responsible? Its user or the company that deployed it?

Waymo and How Google Dealt with the Dilemma

Accountability is of utmost importance for not just risk analysts but even laymen. In fact, when Google’s parent company- Alphabet Inc. was testing its autonomous car called Waymo, they faced this exact perplexing question. They then petitioned the U.S. National Highway Traffic Safety Administration (NHTSA) who later interpreted that in the case of an accident, it would be the self-driving system rather than the human occupant who would be responsible for an accident.

This interpretation essentially meant the software creator was deemed legally responsible which was partly why Waymo was then subsidized considering that Google’s reputation had to be protected in the case of a fatal accident.

Tesla’s Response

However more recently it is interesting to note that Tesla took a different path. While Tesla had an autonomous level of 2 rather than Waymo’s 4, it stated that the driver instead of the manufacturer was liable for an accident. This was used as defence in multiple lawsuits, where Tesla convincingly claimed that their autonomous system was meant to be supervised by the driver.

Hence in the case of an accident, Tesla would show evidence of onboard logs to prove that the individual in the driver seat had neglected or even over relied on the technology, essentially leaving it unsupervised. This defence had proved successful showing that their system was meant to have human oversight. While there is a rise in recent court rulings where Tesla’s defence seemed unsuccessful with the company losing millions, we can see two completely different views from multimillion dollar companies.

What do We Learn from them?

So, we arrive at the climax. We have taken a look into two giants who have dealt with accountability quite differently for their autonomous vehicles. If there is something we can surely understand from them, it’s that humans will be held accountable should they fail to act as oversight over AI.

However, if Artificial Intelligence is at a superior autonomous level, then it is the software creator who is liable. Liability or accountability all comes down to the level of automation. So, on a side note it seems that slower automation has helped quite a few enterprises but if there is something we can learn here it’s that AI if over relied upon will make you liable as a user. But with vigilance, Artificial Intelligence coupled with human intelligence can be unstoppable.

Credits

Cover Image: https://www.linkedin.com/pulse/ai-accountability-whos-responsible-when-your-systems-make-wrong-9ivme

1. https://www.openaccessgovernment.org/article/what-does-it-mean-to-know-something-in-the-age-of-ai/194647/
2. https://www.govtech.com/transportation/waymo-to-begin-testing-its-autonomous-evs-in-minneapolis
3. https://www.tesla.com/en_au/fsd

If you want to read an article about arguably the most heated and common debate topic: “Reading on Paper vs. Pixels”, click on https://ethansliteraryhaven.com/the-great-reading-debate-paper-vs-pixels/.

Do follow https://ethansliteraryhaven.com/ for more such articles.