Is AI the future of the world

This is how artificial intelligence will affect our lives

Yasmin Billeter

Futuristic developments: whether robotic armies, direct brain interfaces or digital brainstorming - Donnacha Daly, the head of the Bachelor's degree in Artificial Intelligence & Machine Learning, AI believes in a lot. That is why he calls for strong regulation.

Mr. Daly, how much has artificial intelligence (AI) already changed our world?

Donnacha Daly: AI is continuously changing our world. But we humans are incredibly good at adapting. Today we use a lot of AI systems, sometimes without even realizing it. For example, when we have a text translated online or use the facial recognition function of our smartphone. Artificial intelligence has arrived in our everyday lives and has changed the business world: ten years ago, the top companies in the USA were large industrialists such as Exxon or General Electric. Today it's Apple, Amazon and Facebook. They all have business models that are underpinned by AI. AI is also helping the world cope with the coronavirus pandemic in many ways. For example, through contact tracing apps that track the chains of infection and thus curb the spread of the virus.

What are the latest developments in AI?

Artificial creativity and AI innovation are increasingly the focus. These are made possible by so-called “Generative Adversarial Networks” (GANs). GANs consist of two neural networks that improve each other. (How this works is explained here using the example of the computer-generated latte art). AI can also bring new knowledge to light. There's a real chance the coronavirus vaccine will be found in part with the help of an AI.

Will there be an intelligence that is smarter than us?

Absolutely! It may take many decades or even centuries, but it will come. A better question is whether an artificial general intelligence has a motive and how it interacts with us. Imagine an intelligence that is relatively far ahead of humanity, much like how we as a species are ahead of ants. We don't really care about ants, except when they nest in our houses. But in general, human motives don't concern ants. Likewise, a superintelligence with a motive might not even consider us humans as part of their motive. Provided we stay outside ...

But you step on ants - accidentally or on purpose ...

That's right, and that's why we need to be careful about developing AI right now. I would never advocate stopping artificial intelligence research, but believe that AI research and exploring the limits of AI should continue at full speed. But the development, commercialization and application of AI should be heavily regulated.

The idea of ​​a robot army is not very far-fetched.

We at the Department of Computer Science take this seriously and work with the Lucerne School of Social Work and the Lucerne School of Design & Art to investigate the social effects of AI and its interactions with ethics, philosophy and even art.

Yet many wonder what happens when AI becomes so intelligent that it becomes a danger.

If used incorrectly, AI can actually pose a threat: for people, cultures and democracies. There is a great risk that AI will be used as a weapon. The idea of ​​a robot army is not very far-fetched. I also believe that if you had fleets of self-organizing, intelligent drones, drones could soon become a problem. Stephen Hawking and Elon Musk warn of this danger. Others like Demis Hassabis, the CEO of Google DeepMind, are more optimistic.

What fundamental rules for AI will we need in the future? Ladan Pooyan-Weihs speaks in an interview, for example, about the importance of digital ethics and calls for the necessary changes to be made.  

I like to compare AI to nuclear energy. It has this transformative potential, for better or for worse. We can achieve great social benefits with nuclear energy. For example, almost 50 percent of electricity in Switzerland is generated by nuclear energy. On the other hand, we have nuclear weapons and have been living with the threat of mutually assured destruction for many decades.

I like to compare AI to nuclear energy. It has this transformative potential, for better or for worse.

How did we as a global community deal with these risks? Through international treaties, global cooperation, independent oversight and very strong regulation. I firmly believe that we need a similar approach to AI.

Which AI regulation laws are in place today?

The European Commission published a white paper on the regulation of AI at the beginning of the year. It says, for example, that machine-influenced decisions must be explainable. This has a huge impact: let's say you have a skin problem. You can go to the doctor, but you can also consult an app. The app would tell them which cluster their skin falls into statistically. According to EU regulations, this diagnosis would not be acceptable as it can only be explained statistically and not medically.

Explainable AI is one of the most important and difficult open problems that need to be properly solved.

I think it is important that we have strong regulation worldwide when it comes to the application of artificial intelligence and its use in decisions that affect human lives.

You respect this rule over the Explainability of AI so as not very useful?

On the contrary! I think explainable AI is one of the most important and difficult open problems that need to be properly solved. Now. Explainable AI will be key to opening Pandora's Box and significantly increasing the use of AI in sensitive application areas such as law, medicine and security. The rule about explainability will also greatly simplify the regulation of AI.

What hurdles does AI currently have in addition to explainability?

Artificial intelligence is terribly hungry for energy. The carbon footprint is AI's dirty secret. We also currently need too much data to train the AI ​​algorithms. If you show a baby 100 pictures of a cat, when it is three years old it will know what a cat is - every cat. An AI has to see 100,000 to 1,000,000 images of a cat before it knows what a cat is. We need AI to work on small data (less, more precise data), not just big data.

Will AI soon compete for our jobs?

Artificial intelligence will do a lot of things, but technology will also create new jobs. Many jobs will change in the time of technological transition. Employees will have to learn new ways to be productive in society. This will also require new social contracts, such as the universal basic income.

What is your prognosis for the future of artificial intelligence?

I explain it with the image of the centaur - a hybrid of horse and human in Greek mythology. Garry Kasparov was the first world chess champion to be beaten by a computer. After his defeat he developed centaur chess: humans and chess computers play together in a team. You can play chess better than humans or computers could ever do on their own.

For the future of AI, this means that people and machines work best in a team. This is then not "Artificial Intelligence", it is "Augmented Intelligence".

Applied to the future of AI, this means that people and machines work best in a team. This is then not "Artificial Intelligence", it is "Augmented Intelligence". Very soon we will see an announcement from Neuralink, Elon Musk's AI society, that will take this concept much further.

The term "advanced intelligence" is used by Ladan Pooyan-Weihs in our interview used as an alternative to the term "artificial intelligence". She says the term "artificial intelligence»Can lead to false expectations in everyday life and stir up unnecessary fears. You understand something else by “Augmented Intelligence” ...

Not really, no, I agree with Ladan Pooyan-Weihs. When we talk about AI today, we generally mean machine learning (ML) and computational intelligence and not neuromorphic computing (a computer modeled on the brain) or even anthropomorphic general intelligence (a computer with human characteristics). The reason for this is that ML is currently the most powerful technology we have for artificial intelligence. And we use it as an aid to human intelligence because it is currently at a sufficiently advanced level to increase our productivity but not advanced enough to replace our ingenuity. So, yes, I agree with Ladan Pooyan-Weihs that we have an “extended intelligence” today.

What will people use AI for in the future?

If you look back ten years, you would ask in a similar way: What will people do with all this computing power? If you had said: someone is going to invent “Angry Birds” and the video game will become the number one application, you would have been declared crazy. But that wasn't crazy, that was creative!

Ten years from now, creative people will be using AI technology in very unexpected ways.

Nonetheless, if you look ten years into the future, you will see enhanced intelligence and a lot of creative people in a similar manner. These will use the technology in very unexpected ways.

Where will we be in ten years?

We will see widespread acceptance of enhancing human cognitive abilities through direct interfaces to the brain. That sounds far-fetched, but innovations such as the iPhone (launched in 2007) and Facebook (launched in 2004) have also changed the world in a very short time.

In spring 2020, the first students of the Bachelor's degree will have Artificial Intelligence & Machine Learning started. How have you fared so far?

I am very sorry for our students who missed many aspects of the first semester due to the corona pandemic: meeting new friends and new experiences. Nevertheless, everyone did an enormous amount of work so that they could at least attend lectures and take exams. I hope that we can carry out the next semesters as planned. Especially when the focus is on project work, creativity and innovation.

AI is of great strategic importance for Central Switzerland, for the national economy and for Switzerland's competitiveness on the international market.

Artificial intelligence is not just about computational intelligence, but also about value creation through application technology and digital brainstorming. This means that we have to work in teams and interact physically.

What are your goals and visions for the course?

My goal is that our students can start their careers with a full set of AI tools, be it in industry or in the public sector. AI is undoubtedly of great strategic importance for Central Switzerland, for the national economy and for Switzerland's competitiveness on the international market.

The original interview was conducted in English in the summer of 2020 and translated for the blog.

Longstanding AI expert: Donnacha Daly is the head of the Artficial Intelligence & Machine Learning bachelor's program and a member of the Algorithmic Business (ABIZ) Research Team. He has worked in various areas of machine learning and signal processing for 20 years. For five years, AI has felt like a technological breakthrough to him, "because the computing power, the algorithms, the data and the high-speed communication are now at a point where the breakthroughs are simply phenomenal". Daly has been active in all of these areas throughout his career.

Educate yourself: Understand AI with the CAS Artificial Intelligence / Artificial Intelligence (AI / KI), apply it innovatively and look behind the hype! The course offers a well-founded introduction to artificial intelligence (AI) algorithms and provides basic knowledge of all important areas of AI.

Get a Bachelor's degree: The pioneering Artificial Intelligence & Machine Learning course focuses on the key technologies of artificial intelligence. Here you can get the tools you need to meet the growing demand in business for skilled workers in the fields of artificial intelligence and robotics.

About the computer science blog: Here you will find tips and news from the world of IT. We portray people and write about technologies that help shape the Lucerne University of Applied Sciences. Subscribe to our blog now and stay informed.

Share post