Generative AI is all the rage

Since the launch of OpenAI’s ChatGPT in November 2022, Generative AI (Gen AI) and Large Language Models (LLMs) have been the talk of the town. And for good reason. Gen AI is already changing how humans and technology interact, and there is much more still to come.

According to our 2023 study, 31 per cent of Scandinavian consumers have already used at least one Gen AI tool. In addition, 24 per cent have heard about Gen AI tools, but have not yet used one. 45 per cent have not heard about them.

Figure 1. Familiarity with Gen AI
_{Which, if any, of the following Generative AI tools are you aware of?
Which, if any, Generative AI tools have you used?}

_{Base: All adults 16-75. Base size: 2023 (n=4,200).
Note: Multiple choice question.}

A large proportion of the people who have used a Gen AI tool have already become regular users. 4 per cent of Scandinavians who have used a Gen AI tool now use it at least once a day, 19 per cent use one at least once a week, and another 18 per cent at least once a month. So, 42 per cent are using one at least monthly.

Figure 2. Frequency of Gen AI usage
_{You mentioned you have used Generative AI (e.g. ChatGPT, DALL-E, etc.). Which of the following best describes your usage?}

_{Base: All adults 16-75 who have used any Generative AI tool and who did not answer “Do not know”. Base size: 2023 (n=1,283).}

Age is a major factor when it comes to awareness and use of Gen AI. We see a clear pattern in Scandinavia: the older you are, the less likely you are to either use Gen AI or be aware of it. 16-17 year-olds are almost ten times more likely to use Gen AI compared to 65-75 year-olds, and over twice as likely as 35-44 year-olds. Similarly, young people also show the smallest difference between awareness and usage. When it comes to Gen AI, young people clearly want to see for themselves, while older age groups appear to be more satisfied with just having heard of Gen AI without actually having used it.

Figure 3. Awareness and use of Gen AI: Age
_{Which, if any, of the following Generative AI tools are you aware of? (The figure shows the respondents who chose at least one tool)
Which, if any, Generative AI tools have you used? (The figure shows the respondents who chose at least one tool)}

_{Base: All adults 16-75. Base size: 16-17, 2023 (n=131) 18-24, 2023 (n=473) 25-34, 2023 (n=790) 35-44, 2023 (n=712) 45-54, 2023 (n=760) 55-64, 2023 (n=689) 65-75, 2023 (n=645).
Note: Multiple choice question.}

More men than women are aware of and have used Gen AI: 38 per cent of male Scandinavian consumers have used Gen AI tools compared to 24 per cent of female consumers.

Figure 4. Awareness and use of Gen AI: Gender
_{Which, if any, of the following Generative AI tools are you aware of? (The figure shows the respondents who chose at least one tool)
Which, if any, Generative AI tools have you used? (The figure shows the respondents who chose at least one tool)}

_{Base: All adults 16-75. Base size: Men, 2023 (n=2,120) women, 2023 (n=2,063)
Note: Multiple choice question.}

Awareness and usage vary between the Scandinavian countries. 64 per cent of Norwegian consumers have heard about Gen AI tools, compared to just over half of consumers in Sweden and Denmark.

Figure 5. Awareness and use of Gen AI: Country
_{Which, if any, of the following Generative AI tools are you aware of? (The figure shows the respondents who chose at least one tool)
Which, if any, Generative AI tools have you used? (The figure shows the respondents who chose at least one tool)}

_{Base: All adults 16-75. Base size: Norway, 2023 (n=1,050) Sweden, 2023 (n=2,003) Denmark, 2023 (n=1,147).
Note: Multiple choice question.}

ChatGPT tops the list

Some earlier forms of Gen AI have been available for several years, but it was largely due to ChatGPT providing an easily accessible chat interface, powered by a very large language model, that enabled Gen AI to have a breakthrough moment in late 2022, surprising even specialists in the field.

ChatGPT tops by far the list of Gen AI tools used by Scandinavian consumers. However, other platforms have also started to extend their reach, in particular Snapchat’s My AI, a chatbot currently available to Snapchatters. My AI, for example, can answer a trivia question, offer advice on the perfect gift for your friend’s birthday, help plan a hiking trip for a long weekend, or suggest what meal to prepare for dinner.¹

Similarly, Microsoft-powered Bing Chat allows users to ask complex questions and obtain comprehensive answers or summarised information. DALL-E2 is an AI tool that can create realistic images and art from a description provided in natural language, and Midjourney is another tool that creates images based on text prompts. GitHub Copilot helps people write code faster through autocomplete-style suggestions. And QuillBot is an AI grammar, writing and paraphrasing tool for improving all kinds of text and ensuring that the author has the appropriate vocabulary, tone and style for any occasion.²

Figure 6. Awareness and use of Gen AI: Brands
_{Which, if any, of the following Generative AI tools are you aware of?
Which, if any, Generative AI tools have you used?}

_{Base: All adults 16-75. Base size: 2023 (n=4,200).
Note: Multiple choice question. Google Bard was not available to users at the time respondents were surveyed. Numbers on use is therefore not included.}

Breaking down the numbers by country, however, there are vast differences not only within Scandinavia, but even more so around the world − especially when it comes to Snapchat’s My AI, which enjoys extensive popularity in some markets while being almost completely absent in others. For example, 40 per cent of Norwegian consumers have currently heard of My AI, but only 22 and 21 per cent of consumers in Sweden and Denmark respectively – and just 6 per cent in Germany and only 1 per cent in Japan.

Figure 7. Awareness of ChatGPT and Snapchat’s ‘My AI’: Country
_{Which, if any, of the following Generative AI tools are you aware of?}

_{Base: All adults 18-75. Base size: Austria, 2023 (n=1,000) Japan, 2023 (n=2,000) Norway, 2023 (n=999) Netherlands, 2023 (n=2,000) Sweden, 2023 (n=1,995) Belgium, 2023 (n=2,000) Hungary, 2023 (n=1,000) United Kingdom, 2023 (n=4,000) Germany, 2023 (n=2,000) Denmark, 2023 (n=1,000) Italy, 2023 (n=2,000) Poland, 2023 (n=2,001).
Note: Multiple choice question.}

Can Gen AI be used for work?

With the huge global interest in Gen AI, there is increasing focus on how it can be used in business. As with all cognitive tools, outcomes depend on how they are used, and the risks of Gen AI have not yet been explored to the same extent as its capabilities. A question is whether business users can trust the outputs from this kind of AI application, and if not, how can trust be achieved?

For several years, AI (machine learning) has been used by companies and public sector organisations to automate operational tasks, transform business functions, analyse vast amounts of data and create seamless digital customer experiences. However, unexpected advances are already being made with Gen AI, and it seems clear that the technology will provide businesses with extraordinary opportunities to improve their products and services. For example, the marginal cost of producing initial versions of knowledge-intensive content, such as IT code, marketing content or creative design, can be reduced almost to zero, and with a lower carbon footprint compared to creation by humans.³

And Scandinavian consumers are eager to use these tools in their work. 32 per cent of those who have already used Gen AI have used it for work-related purposes.

Figure 8. Work-related use of Gen AI
_{Which of the following purposes have you used any Generative AI for?}

_{Base: All adults 16-75 who have used any Generative AI tool and who are in employment. Base size: 2023 (n=939).}

However, there are concerns about the use of Gen AI in the workplace, given the many instances in which it can potentially take over tasks previously performed by humans. 46 per cent of employees across Scandinavia worry that Gen AI will reduce the number of jobs available in the future – and 23 per cent are concerned that it could replace part of their own role in the workplace.

Figure 9. Concerns regarding Gen AI
_{To what extent do you agree or disagree with the following statements?}

_{Base: All adults 16-75 who are in employment and who have heard of AI. Base size: 2023 (n=1,657).}

Trust or no trust?

There are also concerns about the credibility, bias, misunderstanding and misuse of AI.

Current Gen AI models have significant limitations. As noted by Bloomberg, humans are biased, but Gen AI “is even worse”. For example, one Gen AI model generated images of people with darker skin tones 70 per cent of the time for the keyword “fast-food worker,” even though 70 per cent of fast-food workers in the US are white. Other models have also been shown to amplify gender and racial stereotypes.⁴

Another well-known limitation of Gen AI is “hallucination”. This refers to a high-confidence response that is not grounded in the data and can be entirely false or inaccurate. Misuse of Gen AI is another serious problem. Users are already tricking chatbot AI models by using various methods to bypass their built-in constraints. For example, an AI model may not be allowed to tell you how to rob a bank, but if you ask it to write a theatre play about people robbing a bank, it will take you through all the steps. When this happens, the model may not only provide you with false information or inappropriate content, but may also provide you with authentic data that you were not supposed to see. This problem is a matter of creativity in writing prompts, especially at this early stage of development when the technology is far from mature.⁵

Normally, people who use and are familiar with a given technology are more aware of its limitations than non-users. With Gen AI interestingly − and worryingly − the opposite appears to be true. Although users should be aware that Gen AI can hallucinate and produce inaccurate answers, an astonishing one in five users believe that it is always factually accurate (more than double the 8 per cent of those who are only aware of Gen AI but do not use it). However, Scandinavian users are less overconfident than their UK equivalents, where research has found that 43 per cent of users believe that Gen AI is always accurate.⁶

Figure 10. Attitude towards Gen AI: Factual accuracy
_{To what extent do you agree or disagree with the following statement: “Generative AI always produces factually accurate responses”?}

_{Base: All adults 16-75. Base size: Have heard of Gen AI but have not used it, 2023 (n=994) Have used Gen AI, 2023 (n=1,302).}

Gen AI as a creator

It is one thing to use Gen AI to summarise information, find complex answers, automate tasks or improve writing and grammar. Using it for creative purposes is a different ballgame, and we are only just beginning to discover the endless possibilities it offers – and also the many ethical problems arising from it.

The ability of AI to create a convincing (albeit low-quality and greyscale) image of a human face emerged around ten years ago. Since then, image quality has increased, and today almost anything that can be described in words can also be generated as an image. Throughout 2022, social media users experimented with Gen AI platforms and shared their results. We have seen avocado armchairs and photorealistic images of astronauts riding horses on the Moon. Cosmopolitan magazine was the first to publish a cover page created by an AI-based image generation tool, and there has even been a case of a user submitting an AI-generated image to a fine art competition – and winning first prize.⁷

Given that Scandinavians are still in the process of discovering Gen AI technology, it is too early to examine the opportunities and consequences of using it for creative purposes. However, we asked Scandinavian consumers whether they would be less inclined to listen to music if they knew that it was produced using Gen AI.

Figure 11. Attitude toward music created by Gen AI
_{To what extent do you agree or disagree with the following statement: “I would be less inclined to listen to music if I knew it was produced using Generative AI”?}

_{Base: All adults 16-75 who have heard of AI. Base size: 2023 (n=2,295).}

The picture is muddy. Some consumers are in favour of Gen AI, others are sceptical, and a large proportion simply do not yet know. Interestingly 40 per cent of those surveyed are less inclined to listen to music created by Gen AI.

However, various content owners have started to sue AI platforms for the opportunistic use of copyrighted materials⁸ – and some consumers may question the emotional value of content that is essentially created by a computer.

Heaven for students?

While companies are trying to find ways of using Gen AI, and the creative use of Gen AI is already being challenged in some courts of law, students are one of the groups most obviously benefitting from the technology. They now have a brand-new tool for gathering and summarising large amounts of complex information.

Scandinavian students know this already, and in our 2023 survey more than half of all students who have tried Gen AI have used it for their studies.

Figure 12. Study-related use of Gen AI
_{Which of the following purposes have you used any Generative AI for?}

_{Base: All adults 16-75 who have used any Generative AI tool and who are currently a student. Base size: 2023 (n=229).
Note: Multiple choice question.}

Since Gen AI can be an easy shortcut to searching for information or even writing assignments, educational institutions will need to provide guidance to students on its use – and many are already doing so. For example, Aalborg University does not permit the use of Gen AI to create text that is inserted directly into a project report, and Gen AI and similar technologies cannot be used for re-writing text, since spelling and formulation skills are evaluated in written exams. As the university has stated, good scientific practice in general dictates that you “never present ideas or statements that are not your own without clear referencing. This also applies to the use of Gen AI”.⁹

The impact of Gen AI

Gen AI is a tale that is still being written. Less than a year ago, few people had heard about it. Today, innovations in machine learning and the cloud tech stack, coupled with the popularity of publicly-released applications such as ChatGPT and DALL-E2, are already having a huge impact.

While many Scandinavians are experimenting with Gen AI, at school, at home or at work, many are still just thinking about it. In general, young people are more in favour of it (44 per cent of Scandinavians aged 18-24 are positive), but older generations are also warming to Gen AI, for example with 30 per cent of Scandinavians aged 65-75 having a positive attitude towards it.

Figure 13. Positive attitude towards Gen AI: Age
_{On balance, are you positive or negative about Generative AI, or are your views neither positive nor negative? (The figure shows the respondents who chose “Very positive” or “Fairly positive”)}

_{Base: All adults 16-75 aware of any Generative AI tool. Base size: 16-17, 2023 (n=113) 18-24, 2023 (n=406) 25-34, 2023 (n=589) 35-44, 2023 (n=440) 45-54, 2023 (n=357) 55-64, 2023 (n=219) 65-75, 2023 (n=146).}

What is certain is that over time most people will be affected by Gen AI in one way or another, from outsourcing labour and changing professional roles to human-machine interaction, the trustworthiness of digital content and all the ethical concerns that come with it. Scandinavians seem to realise this. 32 per cent of those who have already used Gen AI believe that it will be an integrated part of their daily activities over the next 12 months – and even 14 per cent of Scandinavians who have not yet used Gen AI believe the same.

Figure 14. Integrating Gen AI into everyday life
_{To what extent do you agree or disagree with the following statement: “In 12 months’ time generative AI will be an integrated part of my daily activities”? (The figure shows the respondents who chose “Strongly agree” or “Tend to agree”)}

_{Base: All adults 16-75. Base size: Have heard of Gen AI but not used it, 2023 (n=993) Have used Gen AI, 2023 (n=1,304).}

Some Gen AI tools are free, but there is a charge for others, especially for premium versions with added features or unlimited usage. It is estimated that each Gen AI query costs less than ten US cents, and the cloud companies that provide Gen AI processing are spending billions of dollars on the chips needed for it. It seems likely that businesses will be required to pay some kind of subscription for access to powerful Gen AI tools or to integrate them into the major corporate software suites. The pricing structure could be a fixed monthly charge (say, $10-30 per user per month) or a hybrid pricing model, with a lower fixed monthly payment but an additional usage-based charge.¹⁰

Some form of subscription for personal use of Gen AI might also be charged the future, and our 2023 survey found that many users would be willing to pay for access to Gen AI, particularly those who are currently using it the most.

Figure 15. Willingness to pay for Gen AI: Frequency of use
_{Would you be willing to pay, for a Generative AI tool, which is available at peak times, and gives faster responses?}

_{Base: All adults 16-75. Base size: Minimum weekly, 2023 (n=298) Monthly, 2023 (n=235) Less often, 2023 (n=229) Used once or twice to try, 2023 (518).}

The legal implications of Gen AI will also be interesting to watch. In June 2023 for example, the European Parliament voted in favour of an EU Artificial Intelligence Act, the world’s first comprehensive AI law aiming to ensure safe, transparent, traceable, non-discriminatory and environmentally-friendly AI systems across the Union. In 2023 and 2024, further negotiations will take place between the EU Commission, the Council of Europe and the EU Parliament before bringing the AI Act into law.

Clearly, the European legislators are well aware that with Gen AI humans work with machines to achieve something neither could do independently, but that now is also the time to establish viable methods of accountability, trust and ethical behaviour when it comes to Gen AI – before it gets out of hand.

_{¹ https://help.snapchat.com/hc/en-gb/articles/13266788358932-What-is-My-AI-on-Snapchat-and-how-do-I-use-it-
² https://genai.umich.edu/resources/tools
³ https://www.tomshardware.com/news/ai-driven-creators-are-better-for-the-environment-than-humans-says-new-study
⁴ https://www.bloomberg.com/graphics/2023-generative-ai-bias/
⁵ https://www2.deloitte.com/dk/da/blog/blog-new-tech/2023/blog-jacob-bock-potent-artificial-Intelligence-models-should-be-an-ecosystem-of-technology.html
⁶ https://www2.deloitte.com/uk/en/pages/press-releases/articles/more-than-four-million-people-in-the-uk-have-used-generative-ai-for-work-deloitte.html
⁷ https://www2.deloitte.com/content/dam/Deloitte/us/Documents/deloitte-analytics/us-ai-institute-gen-ai-for-enterprises.pdf
⁸ https://news.artnet.com/art-world/class-action-lawsuit-ai-generators-deviantart-midjourney-stable-diffusion-2246770
⁹ https://www.studerende.aau.dk/praktisk/it/generativ-ai-pa-aau
¹⁰ https://www.washingtonpost.com/technology/2023/06/05/chatgpt-hidden-cost-gpu-compute/}