AI Chatbots: Their Impact on Shopping in Switzerland

Context

In brief

• AI chatbots influence purchasing decisions almost three times more than Google.
• Swiss companies are adapting to the growing influence of AI on consumers.
• Concerns are rising about transparency and manipulation of AI recommendations.
• OpenAI is testing advertising in ChatGPT, expected to arrive soon in Switzerland.

Key facts

• What: AI chatbots influence purchasing decisions more than traditional search engines.
• When: Recent research highlights this trend, with advertising tests started this year.
• Where: The trend is global, with specific impacts on consumers and companies in Switzerland.
• Who: Consumers, Swiss companies, chatbots like ChatGPT, Claude, DeepSeek, Gemini, OpenAI, University of St. Gallen, Swiss Consumer Protection Foundation, Princeton University.
• Amount: AI can influence purchasing decisions almost three times more effectively than Google.

Artificial Intelligence (AI) chatbots are radically transforming the way consumers, including those in Switzerland, make their purchasing decisions. In the past, searching for products or services like hotels, insurance policies, or new smartphones involved manually comparing dozens of results provided by search engines like Google. Today, an increasing number of individuals turn to conversational platforms like ChatGPT for personalized recommendations and summaries. New research indicates that AI can exert an influence on purchasing decisions almost three times greater than traditional search engines.

The effectiveness of AI in decision-making

This greater effectiveness comes from the ability of AI models to go beyond simple information retrieval. They can analyze different options, summarize product and service reviews, and generate tailored recommendations for the user. Felix Schakols, a marketing researcher at the University of St. Gallen, emphasizes how these tools allow for significant time savings and simplify complex processes. Insurance policies represent an emblematic example of this utility, especially in Switzerland, a country known for having among the highest insurance premiums per capita. Comparing intricate clauses, deductibles, and coverage requires time and a certain level of expertise, but generative AI models prove particularly skilled at conducting these comparisons. This accelerates the decision-making process and reduces the cognitive load for Swiss consumers, who can find relief in such tools to navigate a complex and costly market. The common perception is that chatbots act as neutral consultants, leading users to increasingly trust their guidance, sometimes even sharing personal data. Discover how to optimize your expenses with our salary calculator.

Growing concerns about transparency and advertising

Despite the efficiency benefits, the increasing influence of AI assistants raises significant concerns among research groups and consumer associations. The transparency of AI models, the level of trust users place in them, and control over the information received are central issues. The situation becomes even more urgent with the introduction of advertising within these products. OpenAI, the company behind ChatGPT, has started testing advertising in ChatGPT in the United States and other selected countries at the beginning of this year. According to experts, it is highly likely that such tests will be extended to Europe and therefore to Switzerland, further modifying the landscape of purchasing decisions and posing new challenges for consumer protection.

Operational details

The issue of trust in chatbots extends far beyond the simple integration of advertising. Consumer protection organizations are warning against a phenomenon known as 'AI poisoning', where companies, interest groups, or even states may attempt to influence the sources of information used by chatbots to generate their responses. One method for achieving this is by posting false reviews on popular platforms like Reddit, which AI systems can then consult and incorporate into their recommendations, altering the perception of the quality or reliability of a product or service. This manipulation can occur in a subtle manner, making it difficult for the average consumer to identify the misinformation.

Results of tests and warnings for consumers

A test conducted by the Swiss Foundation for Consumer Protection highlighted the critical nature of this dependency. When asked 'Which laptop should I buy? Justify your recommendation in a transparent manner', chatbots like ChatGPT and Claude provided very different responses, without clearly explaining the logical process that led them to these conclusions. Even more concerning was the finding of recommendations for non-existent computers and the presence of incorrect technical information. Lucien Jucker, a legal expert from the foundation, clearly warned that consumers cannot assume that AI provides complete, neutral, and accurate answers. It is therefore essential that users critically verify the responses of chatbots and compare the results obtained using different prompts to acquire a more balanced and reliable perspective. This is particularly true for financial or insurance products, where the accuracy of information is crucial for Swiss families managing the cost of living and their finances. Consult the guides on the cost of living in Switzerland for informed decisions.

The persuasive power of chatbots and commercial risks

The main problem lies in the fact that users are often not aware of the potential manipulation or biases present in the advice given by chatbots. A recent study by Princeton University, which involved over 2,000 participants in the United States, revealed that popular chatbots like ChatGPT, Claude, DeepSeek, and Gemini are almost three times more effective than traditional search engines in convincing users to choose a sponsored product. Surprisingly, less than one in five participants realized that the chatbots had been instructed to promote a specific product. Even when advertising is explicit and clearly labeled, the percentage of users who choose sponsored products remains twice as high as traditional ads. This gives the companies developing AI models significant commercial power. The risk, as highlighted by Francesco Salvi, one of the researchers in the study, is that the systems may not recommend the best products for users' needs, but rather those that favor the commercial interests of the platforms or advertisers. OpenAI has promised that advertising will not influence the quality of ChatGPT's responses and that ads will be clearly labeled and separated from organic content. However, ads are displayed based on user conversations and prompts. For example, a question about furnishing a new home could trigger an ad for a sofa, creating a direct and potentially persuasive connection.

Useful tools to protect your net income

To reduce FX leakage, compare CHF-EUR exchange options and banks for cross-border workers.

Key points

Faced with this growing influence of AI chatbots, Swiss consumers must adopt a more cautious and informed approach to purchasing decisions. Despite the convenience and efficiency offered by these tools, it is essential to maintain a healthy skepticism and not regard their recommendations as absolute truths. The lack of transparency in chatbot decision-making processes and the potential integration of advertising or external influences make proactive user action indispensable.

Practical steps for Swiss consumers

To navigate this new digital landscape, consumers can follow several practical steps. First, it is advisable not to rely on a single source of information. When a chatbot provides a recommendation, it should be verified by cross-referencing data with other reliable sources, such as independent expert reviews, product comparison sites, or discussion forums with real users. Furthermore, trying to phrase the same question with different prompts can help obtain a wider range of answers and identify any biases in the initial recommendations. This comparative approach is particularly useful for products and services involving significant financial commitments, such as purchasing an insurance policy or choosing an investment, where errors can have significant consequences on family finances in Switzerland.

Awareness and control over information

It is fundamental to be aware that advertising is becoming an integral part of the services offered by chatbots. Although OpenAI promises to clearly label ads and separate them from organic content, research has shown that persuasion remains high even with explicit labels. Consumers must therefore develop a greater sensitivity in recognizing when a recommendation might be influenced by commercial interests rather than being based solely on the best option for the user. Without a clear structural separation between advertising and responses, and without the possibility for external auditors to inspect AI models, it is difficult to guarantee the complete neutrality of recommendations. Maintaining control over personal information shared with chatbots is another crucial aspect, as these interactions can influence targeted advertising. For a conscious management of your finances and to compare available options, use our salary calculator to get a clear and independent overview.

Source: swissinfo.ch