Thomas’ Reflections

My original perception of self-driving cars was pessimistic. Certainly, there are great dangers in unleashing new and oft misunderstood technology upon society, especially for those unwilling to partake in this emerging technology.

During this assignment, I have come to appreciate the beneficial nature of self-driving vehicles, contrasting with Matt’s opinion going somewhat in the opposite direction.

Possibly most interesting to me was the discussion on deep learning and how we are yet to fully understand it. Opaque though these AI models may be, what is truly fascinating to me is how these models can be trained thoroughly to understand something but still not entirely grasp their terminal objective in the end, as Miles (2021) discussed.

Whilst it is clear to me that my original stance of caution towards autonomous vehicles wasn’t without merit, there is certainly benefit to helping people who are otherwise incapable of driving to get around. This could restore mobility to those who would feel isolated without it.

At the same time, the dangers of AI are very real, and I think that humanity as a whole needs to be very careful with the roll-out of this technology. As we’ve already seen with tools such as ChatGPT and the infamous teenager-like Bing (powered by GPT-4), the technology is both incredibly powerful and clever, but also very much a work in progress.

Imaging something like Bing’s bad behaviour making its way into a self-driving car, even if not intentional, is certainly grounds for some sort of horror show. Indeed, we could see plenty of movies in the near future complaining about “killer cars”, as some kind of bad schtick that parallels a disturbing trend in society.

However, if we plan properly, it’s abundantly possible to create self-driving vehicles that are safer than human drivers and fulfil good, and where such a dystopian future doesn’t have to be so.

Certainly, there are lots of obstacles, from what we need to do with infrastructure – and we will definitely need to do something about that, probably sooner rather than later – to where liability falls. We know from experience that regulations take a while to catch up with technology – even technologies that are decades old are still not properly regulated, such as cryptocurrency -, so it is important that we start working on these problems sooner rather than later.

New Zealand may be one of the least affected countries for now, but our appetite for technology is not an unknown quantity, so the sooner we start adapting for the oncoming self-driving revolution, the better.

Autonomous vehicles won’t replace human drivers overnight, and there will doubtless be issues along the way in doing so, from stubborn people to genuine moral and safety concerns. If nothing else, we certainly do need more data and more research, but having the public aware of these issues and willing to work on them is an important and crucial step to bringing the wonders, and perhaps horrors, of self-driving cars to the forefront of our society.