The great promise of AI: Your agent on the internet

Advertising, marketing and the internet itself will change as will he economics of, and returns from, providing these things.

Putting the hype to one side, many questions surround the development of AI, particularly who will win and lose as consumer adoption accelerates.

Admittedly, there’s much we still don’t know. Will consumers continue to navigate websites and apps the way they do today, using text and images? In other words, will AI merely supplement what we already do?

Or will consumers adopt AI “agents” that act on their behalf, anticipate their needs, and execute tasks autonomously throughout the day?

A lot of market value has been created among the bottom layer of the AI “stack”, including chip manufacturers like the Netherlands’ ASML and Taiwan Semiconductor, and the hyper-scalers like Nvidia and Microsoft.

This was achieved even amid a backdrop of rising interest rates, and fears about inflation and recession.

But will sustainable wealth be generated at the top layer where the consumer interaction occurs? It’s a big question because substantial gains are indeed possible, as are substantial losses amid those businesses that will be disrupted; so it is tempting to start prospecting.

The early signs suggest AI will be immensely useful and it can already outperform humans at many tasks. When ChatGPT 3 sat the US Multistate Bar Examination, it failed to achieve the 60 per cent pass mark; however, GPT4 achieved a 75 per cent mark and exceeded the human average of 65 per cent.

How soon will artificial intelligence be pre-empting what you want and catering to your needs? Picture: Apptronik

The bigger question of whether AI will be merely supplemental to the consumer experience or more integrated as an agent was perhaps partially answered at a recent Salesforce event in Sydney. The event provided an insight into the capacity of its Einstein AI agent, first launched in 2016, to transform business by enhancing functions such as sales and marketing, improving the customer experience and increasing productivity.

At a practical level, Einstein can point out incorrect classifications of customers in a merchant’s CRM database, write individualised campaign emails to customers based on browsing behaviour and recent purchases, while also autonomously completing CRM database updates – including setting follow-up tasks after simply listening to the salesperson’s phone call with the customer.

This is potentially the start of a more disruptive scenario where AI controls the journey for the merchant and the consumer.

Consider for a moment how influential Google is for desk-based and mobile web browsing, controlling and directing the consumer journey and making or breaking businesses that rely on this traffic.

One can envisage AI agents analysing a consumer’s preferences and needs, and even qualifications, experience and suitability for a job, for a relationship-matching service, for a property purchase or even simply buying an outfit, by directing the buyer to an outcome and rewarding some businesses at the expense of others.

Consider for a moment the selecting, ordering and delivery of the weekly groceries. Using existing technology, albeit clunky, this process can already be completely automated with less waste and cost savings, and with variety, restocking, preferences and needs built in.

Who wins and loses in a world where the consumer isn’t swayed by the images on a box, the marketing on social media or the advertising “real estate” on a Google search page, simply because a machine is doing all the selecting, ordering, paying and delivering? And what happens to text or image-based search in such an environment? Text-based search has been the default for two decades, but will it be 20 years from now?

In this light it becomes increasingly clear why the Microsofts, Googles, Apples and Metas of the world are investing so heavily in AI. The revenue they can earn from controlling traffic, particularly if consumers begin to accelerate their adoption of, and hand over choice to, AI will be immense.

And consumer adoption will accelerate after the recent plateau. The launch of large language models (LLMs) like ChatGPT, Bing and Bard represents the first wave. Another wave will emerge when Apple and Meta launch their AI agents. Indeed, the winners may be those with snug ecosystems like Apple and Google.

In 2023, iPhone users accounted for approximately 18.55 per cent, or 1.36 billion people, out of all 7.3 billion smartphone users worldwide. Apple is expected to launch its first AI solution/agent/experience in 2024. Meanwhile, Google’s Search Generative Experience is being trialled in the US.

It’s still early days, and one is reminded of the launch, out of the beta version, of Google search in 1999 and social media apps in 2009. From a standing start, active WhatsApp users grew to 419 million in the first four years, Facebook users reached 145 million in its first four years, and Twitter hit 54 million over the same period. By 2018 Facebook had over 2.2 billion active users, WhatsApp almost 1.5 billion and today Twitter (X) claims to have a quarter of a billion active daily users.

AI adoption, however, may exceed even those impressive numbers. Twitter, for example, took 1903 days to reach 100 million users, while Facebook reached that milestone in just 1608 days and Instagram hit 100 million users in 854 days. ChatGPT, which is part of the first wave of AI, hit 100 million users in just 61 days.

Whatever the outcome, it is likely that in a world of multiple personal AI ‘‘agents’’ an increase in overall consumer engagement with the internet will occur. While it is certainly possible consumers could individually and personally spend less time on the internet, it is equally possible their agents will produce more sessions. The economic impact of that change remains unknowable.

Some of the shifts in the next two decades will be profound. The reliance on search engines to acquire traffic is one area that will impact merchants and Google simultaneously. Advertising, marketing and the internet’s plumbing will change, and so the economics of, and returns from, providing these things. Meanwhile, consumer data will become stupendously more valuable than it is even now. Knowing how customers behave, where they go and what they want, and when, will become the currency in an increasingly automated world. And perhaps all lower-value goods and platforms such as real estate, car and job platforms become commoditised and undifferentiated.

For investors, the changes will also be immense. Unless they are backed by businesses with prohibitive set-up costs or important patents, for example, buying stocks and sticking them in the bottom draw will become a strategy as relevant to investing as the typewriter is to AI itself.

Roger Montgomery is the founder and CIO of Montgomery Investment Management