The Huawei Mate 60 RS Ultimate, the top-tier smartphone from the Chinese multinational technology corporation’s Mate series, was showcased at the Mobile World Congress 2024 in Barcelona, Spain, on March 9, 2024. It features an octagonal-shaped camera with super macro capabilities and artificial intelligence features that allow users to see details from a distance. (Photo by Joan Cros/NurPhoto via Getty Images)
World Consumer Rights Day is celebrated annually on 15 March “to raise awareness of consumer rights, consumer protection and empowerment”. Since artificial intelligence (AI) has become such a ubiquitous presence in the lives of consumers globally, it is fitting that the theme for 2024 is Fair and Responsible AI for Consumers.
The original intent of AI was to build a neural network — essentially a brain — but what we’re getting is far smarter than that. Geoffrey Hinton, the godfather of AI, said, “We don’t know what AI is doing any more than we know what’s going on in [our] brains”.
Artificial intelligence has a learning potential that is exponentially greater than that of humans. For example, AI systems can learn collectively, drawing insights from diverse sources and benefiting from collaborative learning approaches. In contrast to humans, AI systems know what other AI systems have learnt.
Furthermore, the integrated circuits or chips that provide hardware support for operating AI algorithms are developing rapidly. The most complex chip design has more than 50 billion transistors on a 5-nanometer chip, which is the size of a fingernail.
The rapid progress of AI has also been driven by the unprecedented speed with which customers have embraced it. Traditionally consumers adopted technology at different times for different reasons. Reaching 50 million users took the telephone 75 years, the internet four years, and Facebook two years. But the adoption of AI is nearly automatic because consumers already have the tools, such as smartphones, needed to use it. The AI language model, ChatGPT, took only two months to reach 100 million users.
Invisible innovation
Artificial intelligence can transform stand-alone products into services by providing them with “senses” and “intelligence” that not only follow instructions but also make decisions. Take maps, for example. They used to be paper products, but because of digitalisation and AI they have evolved into services that provide clients with the most effective route while updating in real-time in response to traffic flow. This means that AI is largely invisible to consumers.
These technologies are being integrated behind the scenes into core products and services, especially in back-end services used to improve consumer-facing front-end services. As a result, consumers feel rather than see the effect of AI on their consumption experiences. Better customer service with little to no extra effort or expense is becoming the norm for consumers.
Society’s sentiment is changing in favour of service providers that use AI to do their jobs, according to content marketing speaker Andy Crestodina. As Andrew Carnegie, one of the wealthiest Americans in history, once said, “Capitalism aims to turn luxuries into necessities.” Likewise, where AI initially felt repulsive to some, consumers are starting to see AI integration as a requirement. But the added convenience of AI is presenting us with an ethical dilemma. We must decide between our desire to understand AI and the performance improvements it provides.
Consumers trust their computers or phones but don’t understand how they function. Similarly, AI models may be trusted too quickly. For example, trust that a given exercise routine is best through an AI fitness tracker or why a language-translation earpiece chose a particular word. This unconditional trust in AI outputs is driven by convenience, which tends to prevail in most cases when we use these technologies. If the training data set is incomplete or biased, the AI model may learn incorrect patterns, called AI hallucinations, and misinform consumers. In some instances, AI errors will be easily detected, but they become less obvious when complexity increases.
Generative AI models aim to duplicate existing content in unique ways, which can create significant value. But, garbage in, garbage out. This expression points out that flawed information produces a similar low-quality output. If the data set used to feed the AI model is biased, it might result in the reproduction of biases and power structures.
Some companies are aware of the effect of biases of AI models and are open about how they try to address the errors. For example, chatbots are being designed to stop processing and ask the user for help if it doesn’t know the answer to a particular question. But this is not the case for all AI models.
Trust issues
Important questions must be answered as generative AI models continue to spread throughout all parts of society with minimal or no regulatory control. These models are built on enormous amounts of data acquired from many sources, frequently without the consent of content creators such as artists, writers, journalists, Youtubers, and social media influencers. Not even selfies are safe. Concerns about responsibility, ownership, human rights, privacy and value distribution arise from this. Hopefully, some of these concerns will be addressed on World Consumer Rights Day.
As consumers get spoiled with better services and rely on their products to “do” things, not just “be” things, consumer experience will entail a degree of trust in the AI-enhanced products and companies. Explaining the product features and giving consumers more control will be essential. Google does this with its “incognito mode” and “clear search history”.
Nevertheless, consumers and companies should remain sceptical of AI outputs, as if they were sourced from a smart Wikipedia page with inherently political opinions.
Lisa Esterhuyzen is a junior lecturer in the department of business management at Stellenbosch University. She writes in her personal capacity.
Discussion about this post