Should Scandinavian Countries Build National ChatGPT Versions?

And why Sweden's approach is 'absurd'

Author

Peter Juel Jensen
February 08, 2024

Pioneer AI: Checking the Nordic AI Pulse.
Welcome to 2024’s 5th edition of The Nordic AI Pulse, the weekly newsletter from Pioneer AI, our Nordic-focused AI investment fund. Dive in for a roundup of AI news, deals, and insights covering the Nordic region.

📚 In This Issue:

  1. Should Scandinavian Countries Build National ChatGPT Versions?

  2. Nordic News Roundup

  3. The Million AI Pitch Competition

  4. Cool Tools and Companies

  5. Please “Steal” This Startup Idea

  6. Nordic AI Funding News

  7. Podcast Insights

  8. Upcoming: Nordic AI Events You Shouldn't Miss

Pioneer Insights

Should Scandinavian Countries Build National ChatGPT Versions?

Erik David Johnson

Sweden and The Netherlands are doing it. So, should everyone else do it too?

The “it” is developing a national ChatGPT-like solution. A central question is whether it is worth the money, time, and effort.

Yes, says Erik David Johnson, Chief AI Officer at Delegate. However, the approach is critical.

Erik David Johnson has an impressive AI resume. He has been a leading member of Danish commissions and boards on AI and IT issues, is a popular AI speaker and adviser, and his company was among the select few private entities the EU consulted with during the development process of the recently announced EU AI Act.

I spoke to him about the ifs and hows of countries developing national AI models and their influence on AI companies and start-ups.

The conversation is edited for brevity and clarity.

Many countries have initiated or are considering national, native-language AI projects. Why do you think this is a good idea, considering the time and money such projects often take?

Erik David Johnson: AI will have an increasingly growing influence on our lives. National Large Language Models (LLMs) or ChatGPT-like systems provide several advantages over solely relying on those controlled by OpenAI, Google, etc.

For example, it enables the development and use of systems built on your own morals, ethics, and values. That might not sound like a big issue, but consider how many things are country-dependent and how Danish values differ from those of Silicon Valley.

I can see how something like this would be helpful in an educational setting, but I'm guessing you see many other advantages that countries can gain?

Erik David Johnson: You can build or tailor language models or ChatGPT-like solutions toward specific services and functions while keeping control of sensitive data. For example, a government might want to create a policy Al or a model that interacts with citizens looking for specific information. Similar solutions could be developed for education or healthcare. The list is practically endless.

This approach could also have huge challenges, including costs. Detractors may argue that it makes limited economic sense?

Erik David Johnson: Computing costs and expertise could definitely be issues, as could attracting AI talent, which is in extremely high demand. However, the price depends heavily on the approach. For example, I would go as far as calling the Swedish approach of building an LLM model from scratch absurd from a time and money perspective. Instead, using the best available open-source model and adding your national content and training is much cheaper. The initial investment for a country like Denmark is around 30 million Danish Kroner, with an additional 10 million Danish Kroner annually for maintenance.

So, you are talking about something similar to a super-version of an app from the ChatGPT store?

Erik David Johnson: That might be a way to describe it. OpenAI is the trailblazer, and where their models are now is where everyone else will be in six to eight months, so some of the advanced features wouldn’t be available at first but would come over time.

I can see a host of benefits for public organisations here. But are there any benefits for private companies and start-ups? I find these hard to spot. Also, couldn't a Danish company create such a model and then offer it to public organisations?

Erik David Johnson: Think of how Google Maps essentially decides how you travel – and what would happen if Google suddenly decided to charge for its use. Imagine if we instead had a transportation guidance system for Denmark that was more precise, up-to-date, and free for everyone.

The primary focus of a national model would be the public sector, but there is nothing stopping the government from offering to help companies, including SMEs, by offering the national solution to them and enabling them to pair it with their data.

Also, I would certainly expect contributions from private companies when establishing a national LLM since that is where much of the AI talent resides. However, they should only be included as team members and contributors – possibly in a multi-vendor setup instead of signing over the entire project to a single private vendor.

Finally, I am certainly not advocating solely using the national model. Companies may well see a strong business case in using large language models from Microsoft, OpenAI, or Google, which will continue to be stronger and more feature-rich than their open-source alternatives.

I would love to hear your thoughts on the article and Erik’s points. Let me know on the Pioneer AI LinkedIn page here. 

🎙Nordic News Roundup

  • AI’s impact on productivity and salary depends on your education and profession. [DK] (Grakom)

  • Norway to introduce AI law ‘as soon as possible.’ [NO] (Kode24)

  • Can new AI systems sense a candidate’s emotions during a job interview? [SE] (SVT)

  • Sweden is investing heavily in Silicon Valley. AI is core to the strategy. [SE] (TN)

  • Meet Arc Search, the AI-powered search engine + browser everyone is talking about. [DK] (Computerworld)

  • Thanks to AI, this Norwegian woman can pursue her dream of being a text writer. [NO] (VG)

  • Why companies that invest in AI infrastructure are going to be the long-term winners. [SE] (DI)

  • Danish university is putting AI on the curriculum - and want it to be part of future exams. [DK] [Ritzau]

  • A new survey shows private companies struggle with data security and AI. [SE] (Computer Sweden)

  • Why 2024 will be the year of AI + IoT. [NO] (Digi) (Paywall)

The Million AI Pitch Competition

Pioneer AI is proud to partner with Talent Garden for the 1 million AI pitch competition. At the event, five AI startups will present their solutions and business cases. The winner will receive a one million Danish kroner investment from Pioneer AI and a membership to Talent Garden.

Join us to see how entrepreneurs are using AI to create the businesses of tomorrow.

🛠️ Cool Tools & Companies

  • Nextgen Classroom from Sweden is developing the classroom of the future using AI.

  • Aava AI from Finland uses AI to streamline case research and discovery for lawyers.

  • Subsets from Denmark offers AI-driven user retention solutions.

    Do you know cool tools or companies I should feature in the newsletter? Let me know!

💵Please “Steal” This Startup Idea

We would love to see someone build the following (and if you do, we might invest in you):

Introducing CareerAI, your personal AI-powered career assistant. CareerAI helps streamline the career planning and job search process through conversations, tests, and collaborative research. The platform can analyse your skills, experiences, and career aspirations, offering personalised guidance and job recommendations tailored to your unique profile and helping you with application processes. The platform can integrate additional features such as industry-specific insights, networking opportunities, and professional development resources.

🚀AI Funding News

Select Nordic AI companies that have raised funds in recent months:

Congratulations to all of the founders and teams!

Pioneer AI top 3 deal and funding highlights and insights: 

1: Biila secures €3.5 million. Its platform for optimising car logistics provides two distinct upsides: it improves customer experience while lowering both monetary and environmental costs for companies. This happens thanks to AI’s ability to multiple scenarios and provide rapid optimisation - abilities applicable across all logistics aspects.

2: Combining traditional strengths with AI. Deal numbers indicate that Nordic companies and innovators are successfully combining traditional strengths with AI. For example, deal data from the start of 2020 to today shows healthcare, biotech, energy, and B2B solutions among the leading categories measured by the number of funding deals.

3: AI funding had a good end to 2023. The last quarter of 2023 saw a string of AI deals in the Nordic region. According to Crunchbase data, there were 17 deals in Q4 2023, compared to 14 in Q4 2022. The quarter also saw bigger YoY funding rounds, led by Meltwater’s US$65 million round.

Are there other Nordic companies that should be on this list? Let me know!

🎧 Listen To

  • AI-Podden - technical deep-dive with the inventor of LoRA, including a look at the potential of GFlowNet.

  • AI Revolutionen [DK] - looks at AI and its possible effects on our democratic processes.

  • Lørn.Tech [NO] - examines the ongoing digitisation, including the use of AI.

📅 Training, Meetups & Conferences

Conferences

  • 14 March (2024): Applied AI Nordics: Make AI practical and understandable and inspire how to get value from AI in business. [Copenhagen, Denmark]

  • 14 March (2024): AI-Con 2024: AI in the context of content and marketing. [Oslo, Norway]

  • 17-18 April (2024): AI+: Get the latest updates from the world of artificial intelligence. [Halden, Norway]

  • 24-25 April (2024): Data Innovation Summit: One of the most influential Nordic events on data, analytics and AI. [Stockholm, Sweden]

  • 25 April: Applied AI Nordics: Leading Nordic AI conference. This year’s subject is how to move from opportunistic to strategic AI. [Copenhagen, Denmark]

  • April 27-28: SAIM 2024: Conference for sharing knowledge and results in theory, methodology and applications of Artificial Intelligence, Soft Computing, and Machine Learning. [Copenhagen, Denmark]

  • May 15-17: DevSum 20: International conference exploring the latest trends and technologies within development. [Stockholm, Sweden]

  • May 24-26: ICMLT 9: The Ninth International Conference on Machine Learning Technologies is a forum for exchanges on current challenges in the machine learning field. [Oslo, Norway]

  • June 10-11: SCAI Symposium 2024: The 14th Scandinavian conference on AI (SCAI) that brings together AI researchers and practitioners to present and discuss ongoing AI work and future trends. [Jonkøping, Sweden]

  • June 10-14: HHAI 2024: The third international Hybrid Human Artificial Intelligence conference exploring AI systems that cooperate synergistically, proactively, and purposefully with humans. [Malmø, Sweden]

  • July 18-19: ICAIAE 2024: Scientific conference for exchanging and sharing experiences and results of using AI for education. [Copenhagen, Denmark]

  • October 14-18: Process Mining Conference 2024: Industry and scientific conference on process mining. [Copenhagen, Denmark]

  • October 21-22: Nordic AI Meet: Symposium early-career AI researchers to network, collaborate, exchange ideas, and form a Nordic approach to building AI solutions for the societal good. [Helsinki, Finland]

  • October 21-23: D3A: The Danish Digitalization, Data Science and AI brings together a diverse audience to explore the latest trends and research in AI. [Nyborg, Denmark]

Meetups

Do you know an event, meetup or similar that you think should be on the list? Let me know!

Got an AI story or insight from the Nordics? Share it with me!

Reply to this email or connect on LinkedIn.

Until next time,

PeterHer