Thoughts on AI-driven cars

Thoughts on AI-driven cars

There is a brilliant book on Artificial Intelligence called “You Look Like a Thing and I Love You: How Artificial Intelligence Works and Why It's Making the World a Weirder Place” by Janelle Shane. To be honest, I have not read the book in full yet. However, I have read the excerpt from the book dedicated to problems humanity faces when it tries to incorporate AI into car driving. The excerpt itself I found on the Ideas.Ted. Com and the link will be put below. The reason why we still do not have AI-driven cars (or even vehicles of all kinds) as a universally accepted tool for transportation given human concentration and tendencies to have fatigue made me read this article.

After reading it, I thought that some people – especially not from the industry – tended to exaggerate the level of complexity of AI and the resulting consequences for humanity in case it gets out of our control. The truth is that at least in 2019 – when the article was published – the AI was still not suitable for car driving on a complete human-free base. It proved to be very dangerous to 100% delegate the process to AI. The industry is likely to grow and develop at a space rate but I think that not so many things have changed since the publication of the book and excerpt from it.

Summing up the shortcomings of AI-driven car – by that, I mean full automation stage at which zero involvement is to be expected from a human – are as follows:

1)????AI failed to classify objects properly: sometimes, it was unable to recognize human as human when encountered on the street. Object-bicycle-pedestrian appeared in the identification query which leads to a lethal outcome.

2)????Confused lanes

3)????Classified some objects – like trees – as an obstacle

4)????Failed to see some other cars at all.

Some of those things, I think, may appear funny. However, it should be noted that numerous levels of AI autonomy are possible when it comes to AI-driven cars. Some people say that as of now, it is better not to delegate the entire process of car driving to AI. It is possible that the entire road infrastructure should be altered in case we adopt AI-driven cars. For example, the sober analysis may suggest that some areas delegated to AI-driven cars may appear in cities. This could be a good start for AI-driven cars.?

要查看或添加评论,请登录

社区洞察

其他会员也浏览了