Oleg is the CEO of dxFeed, a market data and financial services firm. Over 20 years of experience in information technology and finance.
Artificial intelligence has been the darling of Wall Street so far in 2023, inspired by the public release of several large language model systems, including Open AI’s ChatGPT, Google’s Bard and Microsoft’s Bing chatbot.
In an otherwise lackluster year of stock market performance, 96% of the S&P 500’s 10% gains as of the end of May have been due to Apple (up 36%), Microsoft (up 38%), Alphabet (up 40%), Amazon (up 45%) and Nvidia (up 175%).
Of the five names, Microsoft and Alphabet have recently launched their own models. Microsoft, Alphabet and Amazon are all involved in the cloud infrastructure on which these large language models run, and Nvidia is directly involved in the “picks and shovels” of semiconductor production, which is crucial to the resource-intensive processing of these systems.
But despite the euphoria surrounding AI’s transformative potential, a dour macroeconomic picture evidenced by the narrowness of this particular stock market rally has many investors wondering whether like the crypto, metaverse and dotcom euphoria that came before it, AI is currently in its own speculative bubble.
The AI Pivot
One sign of bubble-like behavior is the widespread AI pivot that appears to be taking place. As a result of all the attention AI has been receiving, many companies are undergoing a rebrand. AI and its related buzzwords are being injected into marketing copy far and wide, the aim being to find any link, however tangential, to associate your business with artificial intelligence.
In my view, these snap pivots from one hype cycle to the next can make companies appear unmoored and lacking identity in the long run, especially when the bubble breaks and attention moves on to the newest shiny thing.
While some do make genuine attempts to incorporate outside innovations in order to remain relevant, many companies are likely to just be content in paying lip service to AI in the hope that some of its shine rubs off on them. We saw it with dotcom, and we saw it again with crypto.
Currently More Questions Than Answers
In my opinion, this first phase of public experimentation with AI has revealed that not only is it better than most people expected but that the rate at which it’s improving is much faster than we’ve previously experienced with other information technologies.
There are obstacles, too, such as the cost of querying an LLM versus performing a traditional search, which at the time of writing is around 10 times more expensive. With search being touted as low-hanging fruit for AI disruption, the price difference will presumably be an obstacle if something like ChatGPT is going to unseat Google search as our de facto window on the web.
Additionally, new ways of thinking about copyright may be required due to the manner in which these models reappropriate web content. What do ownership, licensing and attribution look like in a world of competing artificial intelligences? Especially when the material they come up with is being used for profit.
Proprietary Data And Domain Specificity
This leads to the topic of proprietary data, which I believe will confer significant advantages to the firms that possess it. In a world where all AI have access to the same information, you can imagine the relative advantage of one system over another gradually disappearing to zero over time.
However, in a world where data is increasingly becoming the most valuable commodity, exclusive access to it could be a significant differentiating factor. For companies like Google and Meta, not only do they have access to a wealth of proprietary data, but they also possess the resources to do something with it.
For other companies with valuable proprietary datasets that lack the expertise to enter the AI race, it may be a matter of licensing that data for use by AI systems and thus being able to monetize it.
Rethinking Your Approach To AI Disruption
This would be my advice to companies observing the AI story from the sidelines, perhaps wondering how these new developments apply to them: Firstly, the bandwagon-jumping approach detailed above is to be avoided.
Secondly, you ought to be encouraging your own people to experiment with these technologies so as to try and determine what kinds of operational efficiencies they can yield.
Thirdly, if you’re running a successful business, regardless of the vertical, there’s almost certainly a degree of domain expertise your company has achieved for it to remain competitive. Try and isolate the areas of expertise your business possesses and start thinking about the data in those domains that you’ve been able to accrue over the years.
Finally, AI as a term is extremely broad. The new systems that have garnered such attention are designed to be much more general in nature, which is an enormous task. Within your own vertical, there are likely to be ways in which your expertise and available data can be harnessed by much more domain-specific algorithms to yield competitive advantages.
Final Thoughts
I believe that the above approach is a much more useful way to think about the coming AI revolution as it goes deeper than just attaching buzzwords like “machine learning” to your process or seeing it as an existential threat. It’s more about understanding where you fit into a burgeoning economy of data and how your own industry advantages can be leveraged in this light. I believe this to be a much more fertile middle ground.
Forbes Business Council is the foremost growth and networking organization for business owners and leaders. Do I qualify?
Read the full article here