Would you trust your life on robots?

Written by Jouko Makkonen and Otto Österman

Sometimes technology fails. There is no news in that. Still, people trust more and more on technology in their everyday life. Sometimes even a bit too much. A good example if this is the use of GPS navigators while driving. People rely on the instructions of navigation systems, although those sometimes lead to detours, sub-optimal or even flawed routes or at the worst case to breaking traffic rules and even dangerous accidents.¹

In cases like above, technology is not the only one that fails. The (bad) decision to follow blindly on the navigation system’s instructions and not to look ahead and re-consider, is made by the driver. Luckily most drivers make better decisions. Human is also a good backup system preventing errors made by technology.

Let’s assume the similar navigation error would happen in a driverless car. Most probably there are also other sensors than GPS that would prevent the car from getting off the road or hitting obstacles, so they should prevent the autonomous car driving off the road. But if the sensors or the software fails, autonomous mobile devices are likely to cause accidents. Recent example of this is when Uber’s self-driving car detected a pedestrian but did not react fast enough.²

So, what is the difference between human and robot decision making? Robot decision making is simple: robot makes decisions in a way that it is programmed to make. It might also have other technical limitations, like the sensors it gets it data from. Positive sides are, that as long as it is not out of power, it will not get tired or let feelings affect its decision making. But as a downside, in case of malfunction, bad decisions can cause problems and dangers.

Human decision making can be weakened by for example tiredness or mental state. The biggest difference compared to robots is the effect of feelings. Feelings might sometimes weaken but they also give totally different aspects on decision making. This is something the robots do not have.

Another thing that robots lack in decision making compared to humans, is ethical thinking. This is also difficult to program, because in ethical issues sometimes even the questions are hard to form, not to mention the outcome. It is more describing to call ethical thinking ”thinking” than just ”decision making”, and it also takes into consideration our feelings, worldview, philosophy, previous experiences and so on. Ethical thinking is also needed in decision making in many fields. Getting back to the previous examples on self-driving cars, a big problem still is how would they make decisions that would need ethical thinking.³

When it comes to rationality, machines beat humans

According to PwC’s Aldous Birchall, who is the head of AI and machine learning, when it comes to making decisions, humans still believe they know best. Machines are expected to be 100% accurate all the time, whereas humans are well under that in their decision-making accuracy. A machine only needs to be more accurate than a human. That’s when machines become useful.⁴

In his TED talk “The key to growth? Race with the machines”, Erik Brynjolfsson, who is a professor at the MIT Sloan School of Management in Cambridge, Massachusetts, argues that “HiPPOs” (highest paid person’s opinions) are usually relying on human-like traits like intuition and biases. In his opinion, this leads to poor decisions. A machine, on the other hand, is stripped of these traits and makes a decision based solely on cold hard facts. This can be beneficial in some situations, but it is not this black and white. Who gets to decide what the machine decides?^{4, 5}

There are many examples of machines making better decisions than humans. One milestone was achieved in 2017 when Google’s AI ‘AlphaGo’ beat the world’s best player of Go, a board game bearing a likeness to chess. It has been said to be the world’s most complicated strategy board game, so this is no small step for AI. A machine is able to predict moves so far into the future, that humans just cannot keep up, even though the grandmaster did win one out of the five games he played with AlphaGo.⁶

There is no question about it - machines can make better decisions than humans. Not all kinds of decisions obviously, only when the data fed to the machine is sound and the decision at hand does not include heavy ethical factors.⁷ Technology is advancing at a very fast pace so we won’t have much time to learn from our mistakes. So let’s just hope nobody creates Skynet.

[1] Driver follows GPS, drives car into a lake:

https://youtu.be/QcrRvQny7mM

[2] Uber’s self driving car detected pedestrian, but did not react fast enough:

https://www.theguardian.com/technology/2018/may/08/ubers-self-driving-car-saw-the-pedestrian-but-didnt-swerve-report

[3] The ethical dilemmas of self-driving cars: 

https://www.theglobeandmail.com/globe-drive/culture/technology/the-ethical-dilemmas-of-self-drivingcars/article37803470/

[4] Why humans must accept robots make better decisions:

https://www.telegraph.co.uk/business/2017/06/25/narcissistic-bosses-biggest-threat-robot-revolution/

[5] The key to growth? Race with the machines (Erik Brynjolfsson TED talk):

https://www.ted.com/talks/erik_brynjolfsson_the_key_to_growth_race_em_with_em_the_machines?language=en&utm_campaign=tedspread&utm_medium=referral&utm_source=tedcomshare

[6] Google’s AlphaGo AI wins three-match series against the world’s best Go player:

https://techcrunch.com/2017/05/24/alphago-beats-planets-best-human-go-player-ke-jie/?guccounter=1

[7] When should machines make decisions?:

https://futureoflife.org/2017/11/30/human-control-principle/?cn-reloaded=1