Nowadays, there is too much data for humans to work on alone, and it is very normal for data analysts to use AI techniques to treat and process these data so it can lead to a faster and more accurate result. But many data analysts and decision-makers still don't trust AI methods or techniques and are reluctant to use them. How can we encourage them to accept or prefer these AI solutions?

For example, if AU gives advice to solve a problem, then decision-makers must trust the results and data analysts must trust the mechanism, so that decision-makers can be confident in the AI work and also data analysts can concentrate their activities on added value.

How can we encourage data analysts and decision-makers to adopt AI?


Not all of the mistrust aimed at AI systems is unjustified, particularly when it comes to neural networks and other such systems that rely on large training data sets. There are a number of high profile cases, facial recognition being one that has often (understandably) received a lot of flak, where improperly configured training data has resulted in skewed and questionable results.

If you want to foster more trust in the systems, it will require better tools for analyzing how they are approaching problems and reaching decisions, as well as determining if there are holes in the training data. It will require a community working in the field that gives a lot more thought and care to how they approach their training data and what unintended biases they may be introducing by forgetting something than has often been displayed presently.

Of course, some people are just distrustful of new technology, but I think it's more interesting to address the more legitimate concerns.

  • $\begingroup$ I think it would be interesting to cite a case where NNs failed catastrophically or were highly "biased". $\endgroup$ – nbro Mar 19 '19 at 20:07
  • $\begingroup$ I really like the fact you mention to use better tools for analyzing on how data analysts and decision makers are approaching problems and reaching decisions. But can you think of may be one tool that can could help in the process? May be in any industry sector? and what do you exactly mean by holes in the training data? is it like a bad(faulty) datasets? $\endgroup$ – Labelle Doriane Mar 19 '19 at 20:35

The surprising fact is, that the acceptance of Artificial Intelligence isn't determined by the technology level but has to do with the motivation of the people. For example, In so training exercises for ambulance so called dummy puppets are used. From a technological standpoint, a dummy is not complicated to understand. It didn't even have an onboard computer. At the same time, such dummy related training is highly realistic, because the puppet is accepted by the human as a legitimate victim. That means, it's part of the game, that the rescue workers have to treat the dummy like a real person, be aware of his emotional state and do what is needed to rescue him.

The same principle can be used for other kind of non-human systems for example Artificial Intelligence robots. Not the robot has to be improved with better technology, but in the roleplay the robot becomes a certain role. This forces the humans to treat the machine not as bot but as a real person.

To answer the question in the headline directly: To nudge decision makers into acceptance of AI work they must be trained in role playing games. That means, it is written in the rules, than the AI is great and then the humans has to use it, no matter if they like it or not.

Robot in the loop The surprising fact is, that in most cases no extra training with dummys or Artificial Intelligence is necessary. Because this is what humans are doing as default. Most games are played with puppets already. In each computer game, simulated forces are used to increase the realism and most rescue training of the firedepartment and scenario based business games are working with non-humans in the loop.

  • $\begingroup$ Thank you for your reply. I gladly appreciate your illustration with the ambulance case using dummy puppets and I totally agree with you on the fact that Artificial Intelligence isn't determined by the technology level but has to do with the motivation of the people, but I really still don't understand what you mean by decision makers trained in role playing games, can you please give more insights to this? $\endgroup$ – Labelle Doriane Mar 19 '19 at 19:43

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.