Product Leader at Conga | Investor | Strategist | Autodidact.
Three weeks after ChatGPT’s release, I published an article, “How ChatGPT Will Forever Change Customer Expectations”—and has it ever. AI has become something practical, useful and real in the eyes of the public.
Now, ChatGPT has its own eyes (plus ears and a mouth) and can browse the internet. The breadth of capabilities is incredible, from improving clinical trial processes to writing localized website content for real estate brokers. This technology is advancing swiftly, and as a result, customer expectations are changing faster than I anticipated.
The narrative is about more than ChatGPT. In the wake of its announcement, a flurry of additional, similar “generative AI” tools have come to the fore and compounded its influence on the world. When ChatGPT was first released, there was slight friction to adoption: You had to sign up for a new website (not a mobile app) to use it.
Now, most of that friction has been removed as “AI” is baked into the tools we already use, such as search, documents, spreadsheets, presentations and CRM systems, including those by providers like Microsoft, Google and Salesforce.
As a result, it’s becoming familiar at unprecedented speed. For consumers, “familiar” drives “expected,” so tech companies must figure out how to meet the expectation to have AI capabilities included or integrated when building applications.
It’s AI Summer
If an “AI Winter” is cooling in the interest and use of this technology—and these, like ice ages, have happened —we are now in “AI Summer,” and the margaritas are flowing. To frame the pace of innovation, here are just a few of the major announcements that have taken place over this period:
• Microsoft integrated generative AI into Bing (February)
• Google released Bard (March)
• Baidu announced ERNIE Bot (March)
• Salesforce announced Einstein GPT (March)
• Alibaba announced Tongyi Qianwen (April)
• Meta released Llama2, making it free for commercial use (July)
Massive investments have accompanied these announcements. Early in the year, Microsoft took the early lead, investing $10 billion in OpenAI and integrating generative AI into its Office 365 and Bing offerings. (OpenAI is now looking at another investment round at three times its then-valuation.) More recently, Amazon invested $4 billion into Anthropic, an OpenAI competitor. Each of these investments carries cross-purchase agreements whereby the AI workloads of the investees are serviced on the investors’ respective cloud (e.g., Azure, AWS).
This activity has also created investor excitement. NVIDIA, a key infrastructure provider for advanced AI workloads, hit $1 trillion in market capitalization. Hugging Face, an AI shop focused on open source, fetched a $4.35 billion valuation—more than double the prior year’s value. This is all amidst a surge of AI-related startup investment.
No longer limited to using proprietary web-based chat interfaces like ChatGPT, the major cloud providers are exposing these capabilities for developers to bake similar functionality into their own end-user applications. Access to start-of-the-art general generative AI is becoming omnipresent and rapidly less expensive—factors that could portend further rapid adoption.
General Generative AI Has Become Winner-Take-Most
In a short period of time, this space has turned into winner-take-most, with users and investments gravitating toward a few leaders. Already, differentiating from the early leaders is hard and only promises to become more difficult due mainly to scale advantages. These are expensive technologies to build and service, after all.
As the winning providers take in more user requests and serve more responses, their tools are becoming multiples stronger versus the next best, creating significant competitive barriers aside from knowledge domains not readily accessible or with limited content on the public internet.
The implications for companies building AI in-house are becoming clearer: Because the big players work so well at relatively low cost, yours will need to do [insert functionality] ten times better or serve a defensible niche. The history of competition shows that it’s difficult to compete with “good enough,” especially in the face of a cost advantage.
Moats in this space could dry up more rapidly, and a few leading players would then take over the castles. In the face of this reality, we’re already seeing several “partnerships” to leverage major cloud provider AI capabilities and more “bring your own model” approaches.
Early Days Of The AI-nternet
I recently heard generative AI compared to the early days of the internet. While it may not be that extreme, we’re only at the beginning of realizing its promise. With general generative AI, it’s simply difficult to understand the full implications when the applications handle such varied use cases and get wider and deeper by the hour.
One difference is the realization of its promise has emerged much faster than the internet did. I suspect a meaningful AI-related announcement will be made between the time this article is submitted and when it’s published—and something still more material shortly afterwards.
So, where is this all going? It’s clear that customer expectations will progress even faster than I originally expected. The power of AI is self-evident and the concerns about it taking our jobs and providing inaccurate information will recede and become more manageable. But we are just at the beginning of the story—generative AI is not yet improving by itself, a logical next step—and I cannot wait to read the rest.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?
Read the full article here
