AI for What?

I don’t believe that current technologies deserve the label AI. In my view, people are confusing the meaning of AI with statistics, automation or brute force computation. Taking WhatsApp or Google as examples, the algorithm seems smart, but basically they run on simple statistic rules. This is the case for many B2C tech products. As a neuroscientist, I have a different perspective on what intelligence really means. If the algorithm needs 300 billion pictures of cats to determine whether it’s a cat, a horse or a cow, I don’t find that particularly intelligent. Artificial intelligence should be able to reason and to be creative and think out of the box. That involves small data rather than deep learning or big data. We’re most likely on the wrong path when we are attributing the nomenclature artificial intelligence to these technologies used nowadays, it’s rather a marketing hype term. A chess board with 13×13 fields is stimulating your brain because you know how to extrapolate known rules to unknown situations. This transfer of know-how, which we also call competence, is still a complete mystery to science. We have no good theory these days how to build such skills into machines. As long as machines will be only executing tasks based on the intelligence of human programmers, we should not label it artificial intelligence. Don’t get me wrong, this is amazing what statistics can solve, also the automation of processes comes with a lot of opportunities to businesses and our society.

Definitely, I think it will be a great world and great future if humans don’t need to work anymore. We should give all the boring and repetitive jobs and tasks to the machines to alleviate humans. Now talking about emotions, I would not necessarily affirm that humans are in every case more sensitive than machines. As an example, not all salespeople are necessarily skilled at sensing emotions or detecting weak signals when conversing with humans. When you try to get to know someone, his/her desires or needs, you have to meet the person multiple times to detect the signals and learn about their character. A little sensor in our blood veins would enable for example salespeople to sense hormone level or other parameters, improving their access to a client’s and true desires, needs and fears. In that way, machines can sense our wishes much better, much faster than human beings, ideally, humans and machines team up together. In video games, characters already simulate facial expressions and mimic human beings. Of course, we cannot touch them yet, but one day that may become possible. Even if we can figure out that those avatars are fake or artificial, it’s probably just a matter of time until those machines look way more like humans. I’m looking forward to entering a world where robots would take over many jobs with more sensitivity than human beings and where machines are sometimes more human than most of our biological colleagues.

Intelligence is in many ways like fire. Back in the days during stone age, our ancestors had the choice to extinguish the fire or to domesticate it. I think they took the right decision for our future. In relation to machines, I see the role of humans to ask the right questions while machines will come up with solutions. The why will be in human hands in order to give directions to AI. We should revolutionize science in general as the current passing of knowledge is very antique and outdated. These days scientists need to read tones of science paper until they are pensioners and may just know 0,1 % of insights form their respective fields of research. If intelligent machines could do that for us, it would push definitely push forward the progression of science and society.

What are the differentiators between humans and animals? Animals are smart, some have self-awareness, many run faster are stronger or much more abundant, however, there seems one aspect where humans are unique: Communication and Collaboration. Human beings significantly outperform all creatures on earth when it comes to collaboration and communication and using tools to archive communications (eg. books). The richness of our facial expressions is key, moreover, we have established a global communication network and infrastructure, ie. the internet connecting all human beings on the planet. We already enhanced social bonds with technology and made them possible at a global scale. Technology enables an improved communication network, making us even more human and connected as a species. It’s delicate to talk about AI ethics whereas some countries are not even agreeing on the adoption of human rights. This means that there are much more pressing ethical issues in the world than artificial intelligence ethics. Technology is a powerful tool that is actually blind for bad or good, only in the hands of a person does it lean towards one or the other. We should hence make sure that people develop human level artificial intelligence and not companies nor authoritarian governments. So far, it seems that the race for AI is mainly pushed forward by governments or large tech corporations. I strongly believe that AI should be developed by the people and for the people. In order to shape the ongoing dialogue about AI at a global scale, the creation of human level AI should be orchestrated from Europe or Switzerland. Just doing ethics and developing even more complex regulatory frameworks without seeking technology leadership may be completely vain. We first need to build the technology to deserve a seat on the table and be able to discuss how to use it. By leaving it to other companies or countries, we would be rather powerless to impact the dialogue on ethics. To wrap it up, being ethical means to develop human level AI for the people and not for governments.

I’m a big fan of the science fiction movie called “HER” starring Joaquin Phoenix. In the scenario, a man falls in love with an artificial intelligence operating system (OS) able to pretend to love him back and keep the user engaged and in a relationship with the OS. If we consider all those people craving for a romantic relationship, not able to find their other half in the physical world, such a loving OS could be a huge market Another very relevant application field for AI is healthcare. When I approach my doctor with green dots on my face, the doctor opens a huge book to determine whether the symptoms are described in that book or not. We need to revolutionize the medical sector in particular, we could delegate very significant research work to machines which would progress the health tech field much faster than humans. We probably also should look out for two or three back up planets, as the one we live on seems not sufficient considering how we treat our world. AI could help to accelerate science and space exploration. I’m a bit skeptical that 10 billion people will in the future live together in peace and harmony. That’s why we need something better than biological human intelligence, we need artificial intelligence. Those are the main applications I would heavily bet on. I think the race for AI will be decided in this decade, why do I believe that? Because of the explosive progression of communication. People with the possibility to access every book and all know-how in the world, can turn into global driving forces. Being able to systematically focus that know-how on the two or three open questions about the principles of intelligence would enable to build human level AI in the not too distant future and pave the way into a new golden age.