The time for companies to embrace AI is now

Tero Ojanperä asserts the current moment is pivotal in deciding who are the global winners of the AI revolution. As he sees it, the significant change of the 2020s is artificial intelligence (AI).

A while ago, we had a discussion with Tero Ojanperä, Chairman and Founding Partner of Silo AI, Europe’s largest private AI lab, and Professor of Practice at Aalto University in Finland, about the use of LUMI and AI disruption. Read more about his thoughts on the matter in the following interview.

According to him, the wave of innovation caused by artificial intelligence will be as large as the wave of innovation launched by the internet in the 21st century or smartphone innovations in the last decade. Supercomputers are key to the revolution.

Q: What is the connection between supercomputers and AI?

In order to use artificial intelligence, it must first be trained. AI has a brain-like neural network and needs to be fed with educational data, such as billions of tokens of text, images, blogs, books, or other sources. This training requires so much computing power that it cannot be done without a supercomputer. In addition to computing resources and data we need applications and so on and a community of actors working together towards the same goal.

Q: What is involved in AI disruption?

What matters is what AI applications are applied to and what their effects are. That is, how, for example, creative work and knowledge work will change when AI is able to do many tasks previously shown to be impossible for the machine. Such tasks include writing, creating strategies, and drawing architectural images.

Professions disappear, and new ones are born. This affects almost every sector of society and everyone as human beings. Change is happening slowly, but if we don’t gear up for it now, we will be behind in the global race.

Q: How should the so-called more traditional industries, such as construction, react to the transformation brought about by AI

Either applications of artificial intelligence are developed by these industries themselves, or development follows from the sidelines. In any case, it is important to be awake enough to try and use applications when they become available.

All in all, this is a major cultural change. Therefore, companies must start shaping the culture so that the latest technology is utilized, but nevertheless not rush headlong into it. This is going to take some time.

Q: What opportunity in the near future do you personally feel most excited about?

Of these concrete tools, my own work has already changed a lot: In writing my recent book AI Revolution, I used AI to search for information and refine it, as well as to write and create illustrations. At first, it seemed difficult, but slowly, I began to realize more and more how to use AI when making a book. I’m excited that things are going to be so much easier.

Q: What should we do as individuals, businesses, and society to protect ourselves from the controversial aspects of change?

Everyone should start using these tools daily, whether ChatGPT or some other medium. Without personal experimentation, one cannot understand change and the possibilities of artificial intelligence.

Companies must create new procedures. Whether you know it or not, like it or not, every company’s software development team already uses these. Therefore, we need to create our own operating models, and at the same time, ensure that the use of artificial intelligence is made more efficient. This change must begin now.

At the societal level, we need to encourage decision-makers to understand the phenomenon better and to actively engage in thinking about what kind of legislation we need and what the consequences will be. For example, is there enough work for everyone, and what if not? We need to start discussing, among other things, the issue of civic pay.

At the time of the interview, LUMI was used to train the University of Turku’s TurkuNLP research group and Silo AI’s Poro language model. In February 2024, the training was completed. According to its authors, the published Poro language model beats the existing open Finnish language models. The Poro model has been published as an open source model, and the data used for its training is also known, according to  a University of Turku press release. In June 2024, together with the University of Turku’s research group TurkuNLP and the European High Performance Language Technologies project, Silo AI launched Viking 13B – a new, larger multilingual model trained on LUMI. This model supports all major Nordic languages, English, and even programming languages, according to Silo AI’s press release.

On July 10th 2024, AMD announced the definite agreement to acquire Silo AI.

The original interview was published in Finnish on the EuroCC Finland’s LinkedIn pages in April 2024. This is an adapted version of this interview.

Author: Mari Uusivirta