If someone recklessly crashes whilst relying on Tesla’s autopilot, should the police be charging the driver or thanking them?
In a morbid, futuristic and kind of macabre way, they should really be thanking them. The result of the crash means that it is less likely to ever happen again to anyone else driving on the same stretch of road. This is because the car itself relies on machine learning, which is constantly collecting data from the driver and reports back every time the driver interferes or corrects a motion during the autopilot.
The year past seen a couple of deaths whilst cars have been under the control of the autopilot and whilst it’s a horrible outcome, it’s also helped to prevent the same thing happening to someone else using the Tesla on the crash site.
Machine learning and automation are the driving factors behind this and they seem destined to be a major part of our future. It’s thought that an autonomous car, like the vehicles Tesla are developing, can essentially replace Taxi drivers and other means of transport. A level 5 autonomous driving level aims to completely replace human involvement when driving a vehicle, with a complete AI system that envisions itself to be of the same competence as a human driver.
There’s a broader lesson to be learned from this, which is about; why taking people out of the loop and relying on machines is a good thing. Machine learning helps detect anomalies and system intruders far faster than it takes humans to do so by analysing the plethora of data logged on a system. Spotting the minimalist changes in a systems behaviour helps to detect and then proceed to deal with the entity.
Whilst machine learning is still very new, the future looks extremely bright for its use, not least the benefits it can bring to enhancing cyber-security. Last year, the Chief Security Researcher at the cybersecurity tech firm, Bitdefender, Alexandru Balan, told TechCrunch: “Machine learning and behavioural analysis is one of the biggest trends in detecting anything and everything these days.”
It works very much like the Tesla automation, in that the system tracks the movements of a human defending against cyber-attacks. It will learn which behaviours look suspicious and how to deal with them, which will in turn lead to more feedback being available to the human as they will have both their own analysis along with the systems. So instead of having one or two humans analysing the system for threats, you now have the machine backing you up providing more commentary and feedback to help deal with these.
Machine learning will help to identify malware intruders on the system which would normally evade human defences and take appropriate actions, be it stopping the malware in its tracks or alerting relevant programs.
It must be stressed, however, that before you throw caution to the wind, whilst having a foundation geared towards the impact of machine learning is most beneficial, a human infrastructure is still required to help triage these threats and maintain the status quo. This should lend itself to becoming a hybrid system, so that even though relying on machines is a good thing, maintaining a human stronghold that can oversee and react accordingly to threats is in place is just as important.