Estonians are known for their tech-savvy approach to digital solutions, with the country often regarded as the world’s most advanced digital society. So, why might Estonians be sceptical of chatbots? The answer is more nuanced than it may seem at first glance. This article will explore the factors that can make Estonians cautious about chatbots – not out of opposition to technology but because they expect them to work seamlessly.
Why Estonians can be reluctant about chatbots
1. Impersonal interactions
One of the most common concerns and criticisms of chatbots is their lack of human touch. Like anyone, Estonians value quick and effective communication, especially with customer service. But when chatbots lack empathy, miss key context, or respond with a generic “I didn’t understand your request,” frustration is inevitable. For straightforward communicators, a chatbot that can’t respond personally or solve the issue quickly will have a hard time winning favour.
This issue isn’t limited to private companies. Estonia has already made headlines with initiatives like Bürokratt, an ambitious national AI assistant platform aimed at facilitating communication between citizens and the government. While promising, Bürokratt’s current implementation has yet to match the efficiency of government chatbots in neighbouring countries. For instance, Latvia’s Rīgas Namu Pārvaldnieks (RNP) chatbot successfully handled around 47,000 conversations during the first five months, and Lithuania’s VRK chatbot answered 1,738 messages per day on average during election week.
I haven’t seen many chatbots on the websites of Estonian public sector organisations or private sector companies. There’s a lot of talk about AI but much less action. At the same time, I feel that Estonians value quick responses to their questions and dislike being put on hold when calling a helpline or navigating homepages in search of the right information. Chatbots can make this process much easier and even provide personalised content instead of just generic responses.
– Kalle Kuusik, Managing Director of Tilde in Estonia.
2. Language limitations
Here’s a big one – language. While chatbots have come a long way in understanding and processing Estonian, it’s still a challenge to achieve the same level of quality and nuance as in widely spoken languages like English. This can lead to occasional awkward translations, misunderstandings, or responses that feel less natural.
That said, commercial solutions are already making notable strides. Large language models like ChatGPT and Copilot have significantly improved their ability to handle Estonian, offering smoother interactions and more accurate responses. While still not perfect, they’ve set a promising benchmark for smaller languages.
To push these advancements even further, projects like TildeLM are stepping in to bridge the gap for underrepresented European languages, including Estonian. Unlike global models, TildeLM is designed as an open resource to support diverse AI applications and serve over 155 million Europeans. As such projects progress, the quality of conversational AI in Estonian – and other smaller languages – will only continue to improve.
3. Early technical limitations
In the early days, chatbots were built on rigid decision-tree models, offering limited responses based on pre-defined questions. Think of it like navigating a flowchart; the bot could respond to simple, routine queries but struggled with anything outside the script. When encountering an out-of-the-ordinary question, these chatbots were about as useful as asking a rock for directions.
Fast forward to today – LLM-powered (large language model) chatbots have changed the game. They’re built with extensive language models that can “understand” context, predict intent, and respond flexibly. This shift has brought a whole new level of responsiveness and complexity to conversational AI, making LLM bots far more capable of handling the diverse needs and requests of users, even in languages like Estonian.
4. High standards for efficiency and excellence
Estonians have high expectations when it comes to technology. After all, this is the country that brought us Skype and has led the way in e-governance, from AI-powered tax systems to automatic transcriptions of court hearings. Technology is expected to work efficiently, save time, and simplify life, not slow things down. A chatbot that can’t keep up or feels cumbersome won’t meet these standards, especially in a society used to digital excellence.
5. Accuracy concerns and fear of misinformation
Another significant concern is the risk of inaccurate responses, often referred to as “hallucinations”. Stories in the media about chatbot mishaps – where bots give incorrect answers or even fabricate information – have heightened scepticism among users and businesses alike. When managers try out other company chatbots that yield inaccurate outputs, it naturally impacts their willingness to adopt the technology within their own organisation.
When implemented thoughtfully with accuracy checks and customised guidelines, chatbots can become highly reliable. Many companies have already seen success with well-designed AI systems, and as these solutions continue to evolve, the potential for trustworthy, effective chatbots will only increase.
Do chatbots deserve a second chance?
Today’s chatbots are a far cry from the limited decision-tree models of the past. Advances in AI, especially in large language models, have brought about remarkable improvements in understanding context, flexibility, and language capabilities. The potential impact of these advancements goes beyond customer service; a recent report by Google and the Estonian government estimates that the widespread adoption of generative AI could contribute as much as € 3 billion, or 8% of Estonia’s GDP, each year.
The report also suggests that as AI technology becomes more ingrained, 61% of the workforce could find themselves collaborating with generative AI, resulting in productivity boosts across various industries. Company-level AI adoption, for example, has been shown to increase productivity by 2–3% on average. With these benefits in mind, it’s clear that embracing these tools could support Estonia’s economy – and, ultimately, enhance users’ experience with tools like chatbots.
So, while past frustrations may still linger, it’s worth remembering that technology evolves. What didn’t work yesterday might just surprise you tomorrow.
High tech, high standards
So, do Estonians really hate chatbots? Probably not – it’s more that they expect technology to meet a high standard of quality and reliability. Early chatbot experiences may not have lived up to these standards, but the technology has come a long way. Chatbots may never replace the warmth of a human conversation, but they’re evolving to handle everyday inquiries efficiently and even fluently in Estonian. With each improvement, chatbots are moving closer to becoming a trusted digital tool in Estonia’s tech-savvy society.
And, who knows? One day, a chatbot in Estonian might just surprise you with how naturally it responds – so much so that you forget it’s a machine at all.
