“Trust in AI is far too low in Sweden”
“The opportunities offered by AI are greater than the risks,” says John Sinderwing, AI expert and corporate consultant in Sweden and the US.
He is concerned that “an exaggerated fear of AI technology” is holding Sweden back when it comes to using profitable machine learning in companies and organisations.
John Sinderwing can be described as both a pioneer and a veteran in the field of AI. He began studying Computer Science and Engineering at KTH in 2016, and graduated with a specialisation in AI machine learning and NLP – Natural Language Processing – in 2021.
He currently works at IT company Entecon as a group leader and consultant in AI solutions in Sweden and the US.
“Taking a six-month AI development break, as certain researchers and business leaders have suggested, would basically mean that all the good players stop their research into AI while all the bad ones continue,” he says.
According to Sinderwing, the biggest AI threat for Europe at the moment is that “the US and China gather far more AI talent, and increase their efficiency through smart robot implementations in people’s everyday working lives.”
“If AI starts replacing human jobs to an increasing extent and we don’t keep up with developments at even the basic level in different kinds of operations, what will happen to the job market in Sweden and the EU?”
He describes the main benefits of using AI for companies and organisations as a massive streamlining of fact gathering and communication, along with lightning-fast calculation and compilation of huge amounts of information.
“Sweden is very clearly already behind in developments. I’m concerned about all the opportunities we are missing out on when it comes to streamlining in the corporate world. From a consultant perspective, I’m frustrated to see that trust in AI is so low in Sweden,” says Sinderwing.
“In the States, for instance, companies save hundreds of thousands of dollars every year by utilising available AI technology for logistics and administration tasks. There, virtually all private companies currently have large data teams, while most companies in Sweden have not even begun to see opportunities yet.”
Björkman Nilsson also emphasises that at lectures and conferences both in Sweden and abroad, he “virtually never meets AI experts who perceive machine learning as a threat to society.”
“Most of them have no qualms about the actual research development. Everyone does agree, however, that we need to keep an eye on the digital lobbying campaigns where robots will increasingly be used in the future,” he says.
“Robot trolls are already influencing politics through lobbying campaigns. But it’s important to remember that these campaigns are essentially run by humans, not machines.”
He dismisses theories that AI could start taking autonomous action, such as independently surfing the web and updating information, in order ultimately to “sacrifice the best interests of humans for its own gain”.
“For me, that’s still science fiction. I’m convinced that such a scenario is still very far off.”
Sinderwing has followed the evolution of GPT close-up from version 1 to version 4, and says that in each stage the focus has been on “bringing in more data into larger models” of the tool.
“The capacity is reaching a limit. What ChatGPT does is build texts using the words that are most likely to suit a context. A lot of people have cried wolf in earlier versions, and for me the development of ChatGPT is like a snowball that’s allowed to roll down a hill: while the structure grows, the basic principle inside is still the same.”
He sees an unexpected contrast in the fact that Swedish companies are slow to adopt the everyday use of AI, while the nation tops global lists of successful new innovation companies.
“It feels like the early 2000s, when a lot of companies and organisations decided not to build their own websites as they were ‘not needed’. The competition will take market share. If we don’t get on the AI train soon, we’ll miss it.”
Text: Katarina Ahlfort
Photo: Magnus Glans