Where could Generative AI’s impact be most clearly felt in the retail sector?
Roy Amara, a researcher and computer scientist at the Stanford Research Institute, now called SRI International, coined the so-called Amara’s Law in the 1960s: “We overestimate the impact of technology in the short term and underestimate the effect in the long run.” In many ways, this has been the impact of artificial intelligence and its impact on society, the economy, and to a degree the world of retail.
In the last year, we have seen the press fixated on the emergence of the AI chatbot, with the most notable example being Open AI’s ChatGPT, but we also have Google’s Bard and others. However, although many of us may have tested out an AI Chatbot, its immediate impact has been relatively limited. The Economist, in an article on 16th July 2023, made an analogy between the impact of AI on the economy and the introduction of the tractor to agriculture. The tractor was invented in the nineteenth century, but by the 1920s, only 4 percent of American farms had one, and less than half by the 1950s. The impact of the tractor was ultimately large, but its impact took time to feed through into the wider economy. The question then is: what will be the short, medium, and long-term impacts of AI on the world of retail?
Although a survey from McKinsey on 1st August 2023, heralded 2023 as the ‘breakout year’ for Generative AI applications, and indeed many C-Suite executives are aware of these applications and are indeed applying them in limited fields, AI adoption has remained relatively steady. McKinsey’s study found that fifty-five percent of respondents said their organisations had adopted AI, but less than a third said this was beyond one business function. Customer service operations is cited as one area where the impact of AI is gaining traction among those who responded to the McKinsey survey. We are maybe not yet experiencing an AI revolution but more an evolution in use.
Many thought that personal AI retail assistants such as Amazon’s Alexa could end the retail market and change the way consumers buy products. However, although widely used, their consumer impact has largely been limited to playing music and obtaining basic information, such as information on the weather. Indeed, the way consumers search for products online to buy has been surprisingly consistent since the late 1990s. The rise of online shopping over the past two decades has fundamentally changed the world of retail, but the way one searches for products online, in the main, is fundamentally the same as that of the late nineties, apart from maybe no longer using a dial-up connection to the internet, and the key gatekeeper to that world remains the Google search box.
Google achieved market leader status in most countries in the online search market in the late nineties and early noughties. Most online searches for products and services are conducted via Google because its algorithm is perceived as the most accurate search algorithm in the market. It subsequently built itself into the fourth-biggest company in the world based on keyword advertising. The way search looks to consumers, the way Google makes money, and its market have fundamentally stayed the same for twenty years. One searches for a product term, maybe a branded term, and a list of search results appears as the ‘natural search results’. Alongside those natural search results whether above or to the right side of the results, appear keyword advertisements triggered by the search term via the Google Adwords service. However, will AI chatbots change that twenty-year constant? Well, maybe, and what impact does it have on trade mark law and practice?
Microsoft has invested US$10 billion in OpenAI, obtaining a 46 percent stake in the company. Since February of this year, AI chatbots of a similar nature as ChatGPT have been incorporated into Bing’s online search service. Google immediately countered Microsoft’s move with the launch of its own AI chatbot Bard. Why is Microsoft investing so heavily in AI chatbots and targeting the search market, and why is Google reacting? In short, because the online search market is so lucrative, it is also to a degree the gate to the online retail market. The press has tended to concentrate its comments on AI chatbots on the creation of new works of art or out-of-control weapons systems, but their greatest long-term impact, as the Economist has theorised, could be on the online search market.
However, AI chatbots are not perfect. AI chatbots are only as good as the data on which they are trained, and they tend to be trained on information available on the web, which, as we all will agree, is not perfect! Although there are training loops built into these systems to constantly improve them over time, they do make errors or have so-called hallucinations. If they do not know an answer, as with humans, they have a tendency to make things up, and this is where it gets interesting for trade mark attorneys.
If AI chatbots take only twenty percent of the online search market, their impact will be significant. Changing the way online search looks and potentially raising old IP problems in new forms and completely new IP issues. Also, trade mark law is fundamentally based on human error. It seems likely that the hallucinations of AI chatbots will get less frequent with time as they learn, but they will still have flaws, not unlike humans. Therefore, will there simply be a new player in the world of online search with all the same imperfections as humans? Words such as ‘confusion’ and ‘imperfect recollection’ may not be so redundant in the world of online search and trade mark law with the rise of AI chatbots. With fuzzy searches, will the AI chatbot distinguish between a NIKE and NIKF of this world, as Justice Richard Arnold postulated in an online discussion on AI run by the University of Maastricht back in 2021?
Can AI chatbots be manipulated? If hallucinations have become part of the present AI chatbot world, could AI chatbots be misled into pointing a consumer towards counterfeit products? We have already seen the rise of a search optimisation industry with regards traditional online search; could that develop with regards to AI chatbots? The press has recently reported the emergence of the so-called FraudGPT product, a product sold on the dark web which facilitates cyberattacks. Could similar Generative AI chatbots be developed to mislead and confuse your humble retail chatbot? Could AI chatbots compete with one another?
Also, the look and feel of the online search experience could change. It seems likely that keyword advertising could develop with AI chatbots as with conventional text search. A competitor could bid on a search term that triggers an AI chatbot to make a certain product recommendation. This raises the issue of visibility of such bidding to the consumer. Further, the fundamental look of a search could change. At present, text searching produces lists of websites of interest ranked by the search algorithm by relevance. However, AI chatbots are much more likely to provide a more limited number of search results.
It is still open to debate what the impact of Generative AI chatbots will have on online search systems, but their impact could be large. The fact that the major search companies are investing billions into such products and applying them to online searches implies they at least see the threats and opportunities of such systems.
This article was prepared by HGF Partner and Trade Mark Attorney Lee Curtis