Since we are going to talk about robotics and ethics, I find it necessary to review a little bit what robotics is.
In the movies collective imaginary, robotics is typically the self-driving vehicle that will drive on its own, it's typically a more or less humanoid robot.
Robot, or partner, which will be able to hold an intelligent conversation, perform very complex tasks.
Actually, for now, and in a probably distant future, robotics is far from that, because those humanoid robots which are remarkable research tools, are very far from being ready.
There are a lot of scientific breakthroughs to achieve before deploying them for real, and that's without mentioning energy consumption problems.
Robotics is something else.
It is true that in the last few years, robots have overcome the obstacles that existed between the industrial robot, which was in a locked cage, humans were kept apart and there was no room for humans in the robots world.
The obstacles for that kind of robotics are being overcome. Robots are going to get closer to humans.
It's already the case.  Take that hoover robot for example, everybody knows it, well that robot is not in a cage, it's close.
That's the great tendency in today's robotics. Robots are coming closer to humans.  But that proximity involves risks and legal issues, or liability issues, and ethical problems.
Let's get back to the self-driving vehicle. First, let's be a bit honest.
We are not quite ready to have an autonomous vehicle capable of working under any circumstances, everywhere.
For the moment,  don't take a self-driving vehicle for driving roads  covered with snow in the Cantal, you would encounter big problems with those vehicles' decision-making.
However, in an especially  well-structured environment, if it knows the place well,it is indeed possible to use  a self-driving vehicle. Which of courses raises a few ethical issues.
A rather classical  problem would be: such a vehicle will  inevitably find itself in a situation where it will have to choose between 2 tragic choices
From a mechanical point of view, it will have to brake because there is an obstacle on the road, and while braking, unfortunately it will have to go towards one or several pedestrians, or head for a barrier.
Each of those options involves potentially lethal damages for the objects hitting, so either the pedestrian or the driver of the vehicle.
So what to choose?
Who must be saved?
The driver or the pedestrian ?
Some of my colleagues assert that since  the driver has paid for enjoying increased security inside his vehicle, then the driver must be saved at all costs.
Let's look at the problem  from another angle: now we're not talking about just one pedestrian versus one driver, we're talking about a group of people, a group of X people and one driver in the same situation.
What to do?
From which value of X do we start to assume that the pedestrians must be saved, rather than the driver?
That's a typical ethical problem for which we, as scientific experts, must display the available possibilities. Here is what can happen with the benefits and drawbacks.
Of course it does not fall upon us scientists, but simply as citizens and the whole society, to make choices to decide
what is acceptable,what is not, and what must be done if something unacceptable happens.
How do we take action  against people that would disrespect the rules established by society
So robots are getting closer to humans.
And indeed, one of the possibilities would be to use robots to help.
To help and potentially also to observe man's behavior under certain circumstances and to manage emergency situations.
Those are two possibilities. The industrial world has started  using those assistance robots.  They are called cobots. Which means systems where there is both human and robot.
The robot will handle  the most arduous tasks, to lift heavy loads, for example. However, man has the gesture expertise that robots will never achieve, but has difficulties regarding the task's applicability.
So having those two systems collaborate is just common sense, it seems to me. Within industrial robotics framework, from a legal perspective, the situation is quite clear.
There is the machine directive for example, which would need to be adapted to our new forms of robotics, but all in all, the legal grounds more or less exist.
What would happen if there was an accident?
All this is pretty well figured out. So from a legal perspective and regarding the risks,  we are walking on a ground that has already been cleared out.
On the contrary, the observational  aspect is not as easy  and poses ethical problems, because the machine helping the human will also be able to give information on the way he provides and the way he performs his work
That is what is called observational, the system observes human behaviour.
So, of course, a lot of arguments can be provided to assert that it is a good thing, that performance indicators which are revered by managers will be very useful to improve the company's performance.
At the same time, those  indicators are often created with a machine that does not fully understand the context in which the task is carried out. So there can be true reasons for performing a task more slowly than normal because there was an additional complexity factor that indicators cannot measure.
So there is a slight problem with having people evaluated by machines that do not necessarily have the capacity to comprehend the situation. Now, having people observed by machines is  a problem.
Does the technician  performing the task  will consent to  being observed by a machine?
What do we tell that technician about what the machine does, about what it measures?
What is the purpose of collecting the data? Is it anonymized? Is it collected for purely statistical ends? Is it going to leave the company?
Those are ethical problems that will have to be dealt with someday.
We talked about industrial environments, so cobotics. So now let's talk about personal environments.
For example, what you see behind us is the simulation of an apartment for elderly people.
Elderly people can have disability issues for example, they need help to move around so that they can  keep being self-sufficient. At the same time, those assistance devices, which are very present in this room, are often very unobtrusive. Here there are approximately 20 robots functioning all the time.
Those assistance devices  have another purpose, which is directed  towards the medical community. The aim is to assess those people's health condition, they are a high-risk population, in case they fall and that could have  dramatic consequences.
Being able to observe their behavior and report emergency situations. Being able to detect an emerging illness, because when walking with the help of an assistance device, the machine will be able to measure an anomaly, and that will draw the attention of doctors:
"Here there was an anomaly.  There may be an emerging illness.   Diagnosing that person would be appropriate."
So there are undeniable benefits.  But there are also ethical problems.
What is to be done with those devices that are often powerful machines?
You need a considerably powerful machine to be able to lift a person.  This means that power could be used to hurt the elderly person. That increased risk  poses the problem of acceptance.
How far are we going to allow machines to take care   of a person's mobility knowing that those machines will inevitably show defects someday,  which could lead to dramatic situations.
Again, we come upon the robotics and cobotics problem.
That assistant used to observe individuals. That's well-intentioned.
It's used to measure health condition parameters.
But, of course, that data can hold a huge economic interest for people from the outside who will use elderly people's weaknesses to try and sell them life insurance, remote assistance, any solutions you could think of that still need to demonstrate their efficiency but that hold a considerable economic interest for companies.
So, how much observation can we tolerate? What information will be given to the subject that is equipped with those machines or that uses one of his daily life objects to assess his condition? What do we tell him is being observed?
Using simple enough words so that it can be  understood by every users.  That poses a data security problem. Yes, that data can be anonymised, Daniel Le Metayer talks about that much better than I do. The data can be encrypted. That's what we do systematically.
But despite all that, that data has to be handed over at some point. Handed over to whom? In the chosen framework of a doctor-client relationship. So just to the doctor. But that data holds a wider interest for the medical community since a lot of things can be observed on a wide range of individuals. That could help improve some treatments based on the analysis of big data.
So there is this kind of paradox where data should be as confidential as possible, and at the same time, it should be used at least by very experienced people so that medicine could benefit from it.
Then, there is transmission security.
So first, let's demythologize a bit the very reliably hyper-connected world we keep hearing about.
That's far from the truth, and you don't need to move very far  from where I live, that's 40 kilometers from here, to understand that you almost can't use your mobile there, so transmission is absolutely necessary because it is critical.
Imagine an elderly person that falls, that information needs to be transmitted to get to save that person. But we could be in the situation where the classical information grid does not work.
That also means that we need to invent alternative solutions to the WiFi network and the Hertzian network  that will allow transmitting that information. That's the case in the room behind us where we systematically use infrared transmissions because we know  that some elderly people do not have Internet, so there's no WiFi, or maybe the person turned it off.
So we need to find alternative information transmission  solutions, while guaranteeing safety, that an outsider cannot intercept That information  to use them in a deceptive way.
Those are problems that we, as scientific experts, are able to foresee, so to predict the possible futures with benefits  and drawbacks, but it falls to the whole society, the legislator, the citizens to decide what will be acceptable,and what will not.