Arkiver

Nyheder

Aarhus focuses on AI safety research

ai-sikkerhed

From a waterfront office in Aarhus a dedicated community of AI Safety enthusiasts and experts are now trying to tackle the task of how humanity will survive the rise of superintelligent machines

”Let’s save the World together!”

These were the dramatic words ending Erik B. Jacobsen’s talk on AI-safety at Dokk1 during this year’s Internet Week Denmark in the city of Aarhus. He is under the impression that Artificial Intelligence remains one of the greatest challenges humanity is facing right now – greater than climate change and the danger of nuclear bombs.

With this motivation, Erik B. Jacobsen, who has a master’s in chemistry and nanoscience from Aarhus University, has since his talk in May been working on establishing ‘AI Safety Aarhus’ in a small office space at Aarhus University’s Navitas building where around 50 tech-startups also thrive in local incubator space Incuba.

“At this point it’s all about making people realize how incredibly important the field of AI Safety actually is. Right now we are meeting with relevant stakeholders, running study groups on the latest scientific papers and organize public lectures like the one during Internet Week Denmark where 200 people came. In the longer run it is our goal to be able to contribute with fundamental research in AI Safety”, Erik B. Jacobsen explains.

The idea is to create a research- and knowledge center for the safety aspects of Artificial Intelligence in Aarhus. If this succeeds, the working group joins an exclusive club of about a handful of research units that currently work on this globally. And in Aarhus they are working eagerly to raise suficient funding to get there.

Be careful what you wish for

When thinking of the threat from Artificial Intelligence, it is not an army of Termanitor robots you should picture, rather an extremely smart computer.

“Imagine a computer, that knows more than all of humanity combined. That’s cool to have, as long as it’s doing stuff we want it to do, but if it does something we don’t want and we can’t stop it because it’s smarter than us, then we really have a problem”, Erik B. Jacobsen explains.

These concerns have also reached the United Nations and The White House recently, both now putting focus on longterm risks and benefits with the development of Artificial Intelligence.

One dilemma that is being discussed is how to teach Artificial Intelligence to understand what we humans actually mean – not just take what we say in its literal sense. If we want the AI to solve the climate challenge for us for instance, it is not a good solution to just evaporate all of humanity in the process.

Eliezer Yudkowsky, a reknown AI theorist within the field of AI puts it like this:

“The AI does not hate you, nor does it love you, but you are made out of atoms which it can use for something else.”

Erik B. Jacobsen gives us a deeper insight into the challenge of AI Safety in this talk from Internet Week Denmark here: https://video.twentythree.net/artificial-intelligence-meeting-aarhus

A three-step process in Aarhus

Research in AI Safety balances a long way between philosophy and technology. What is right and wrong, and how do you teach that to a programmed brain?

The AI Safety Aarhus initiative is the first step in a process of three phases, that eventually will lead to a research- and knowledge center for Artificial Intelligence in Aarhus, where the people involved hopefully will come up with answers to how we can ensure humanity does not lose control over AI.

-“We now have a place, where we can facilitate meetings and lectures, and where we can begin the fund raising process. Next step will be to gather funds to run a research project on AI Safety over 2-5 years.

Ultimately the third step is to gather enough funding to create a fundamental research team of around 10 people that can contribute with worldclass research on the topic, Erik B. Jacobsen says.

Creating a cluster effect for Aarhus

Even though the research centre has no plans to profit directly on their work, but instead run it on a combination of public and private fundings, it is not unlikely that it can help to create jobs in the city of Aarhus (where companies like Google and Uber are already placed).

“It’s not about making money, this is about saving humanity. It’s as simple as that. With that being said, it does of course also make a lot of sense for a city like Aarhus with regards to creating jobs” Erik B. Jacobsen explains, adding:

“The region has many very talented people within the IT sector, and also startups working within this particular field – I was already recently contacted by a startup from Silicon Valley, that was looking for developers here in the city. It could be huge win for the whole region to position itself on the global market within this field.”

With knowledge so specialized gathered in one place; it will naturally affect both the university and the companies in the area positively. But first the funding needs to get in place, to reach the second phase of development.

Erik B. Jacobsen urges everyone who wish to contribute to the project, to join the Facebook-group ‘AI Safety Aarhus’ or to follow the news portal ‘Reality Check Denmark’.

Danish version of the article: trendsonline.dk/2016/08/08/aarhus-skal-vaere-centrum-for-ai-sikkerhed/

erik3_FotoMathiasSvold

Skriv et svar