Future skills for a brave new world

Understanding how artificial intelligence (AI) works is a key skill – now and for tomorrow – says leading computer science and information technology professor Matti Tedre.

Reading the headlines about artificial intelligence (AI) could easily lead you to believe it’s a unified force that’s going to take over the world. To cite just one example, recent news about ChatGPT has many people thinking the AI chatbot will replace any human being who creates copy, such as a writer or journalist.

Yet beyond the headlines and common misconceptions, there’s another story, says Matti Tedre, a leading international information technology expert and computer science professor at the School of Computing, University of Eastern Finland.

“AI refers to specific technologies. There’s not one AI that does it all,” says Tedre.

Simply put, AI means using computers to do things that traditionally require human intelligence.

“AI is a concept that’s been around for 70 years. In the same way that digitalisation has changed our lives over the past two decades, it’s beneficial to understand what AI technologies can and cannot do,” says Tedre. “This is important as the biggest change is the amount of data that’s available to us now,” he says.

Tedre is currently in charge of the Academy of Finland’s Generation AI research project that aims to strengthen the ability of children and young people to understand technologies based on AI.

As AI impacts our daily lives and how the world appears to us by feeding us content on social media, choosing answers given by search engines, and customising targeted ads, knowing how AI works in three key areas is crucial for people of all ages.

Understanding data flows

“First, it’s important to understand information flows, who controls them and where and how they were created,” he says.

A good example is a technology like ChatGPT developed by California-based OpenAI.

“This is super important because ChatGPT is based on information that represents only about one per cent of the Internet. That’s a very narrow, really privileged segment of society and their systems, attitudes and values,” he says. “The voices of a lot of people are not represented.”

For example, African cultural and economic realities are under-represented, which means ChatGPT may have Western cultural and ideological biases. Its data sets and algorithms have little relevance to the reality of life in Africa.

Managing data flows

“Second, it’s important to understand data-driven automation and data agency. This is about volition and capacity in the digital world, as well as active control,” says Tedre.

Data agency extends the concept of data literacy by emphasising people’s ability to not only understand data, but also to actively control and manipulate information flow and to use it wisely and ethically, he explains.

The easiest way to learn difficult things is by trying and doing it yourself. This also applies to artificial intelligence.

Here, it’s useful to understand what information is collected about you and what it’s used for.

For example, in one of Tedre’s projects, students learn how to create AI-made systems that they control and manipulate. This is just as applicable to adult learning and continuing education – if you understand how to create and manipulate the tools, you can make them work for you.

Seeing the bigger picture

“Third, understanding what computers can and cannot do is very important,” says Tedre. “We think that, to do something, the computer needs to understand it but that’s not the case,” he says.

“Computers also don’t understand significance,” says Tedre. “That means that they don’t understand the societal importance of something or its relevance to the big picture. Neither do they understand communities and moods, building relationships or trust.”

One of the biggest problems in AI is algorithmic bias. A good example of this is credited to Tedre’s Generation AI project colleague, university researcher Henriikka Vartiainen: When AI is asked to create a picture of a prime minister, it offers up an image of a grey-haired man.

As the majority of prime ministers in history have been middle-aged men, the algorithm doesn’t know how to correct the bias and assumes this is accurate. Case in point, Finland’s former prime minister Sanna Marin is a young woman with brown hair.

Future world

Looking forward, what will Europe look like in the 2040s with regards to AI? If understanding AI is an essential competency, what capabilities will not be needed in the workforce?

“In terms of the future of jobs, the warning signs of looming automation or that a job can be outsourced are those occupations with predictable outcomes or routines. These are often tasks that a machine can learn. Also, any job that can be learned by watching YouTube videos over and over again will likely become redundant,” says Tedre.

“One job that is going to be heavily outsourced to AI is proofreading and language-check services. We used to pay a lot to many proofreading services to check the final language in PhD theses and PhD students’ articles, but that’s quickly coming to an end. Language problems typical of Finns (articles, prepositions, word order, Finglishisms) are easily fixed by large language models,” says Tedre.

As for Europe’s position in the global AI field, Tedre says that depends on several factors.

“The three big AI players right now are the US, Europe and China,” he says.

“The US is characterised by innovation-driven entrepreneurship with less regulation, Europe is regulating IT in innovation, while China’s model is a government-led Big Brother. The battle is between three different approaches and three different models of an information society. It remains to be seen which one wins and takes all the talent,” says Tedre.

When it comes to AI, knowledge about how it works is power – whether for an individual, an institution or a nation.