The AI Gold Rush: Are You Digging in the Right Spot?
Oh, the allure of AI! It’s like the gold rush of the 21st century. Every budding entrepreneur with a bit of coding knowledge and a dream is looking to strike it rich in this promising landscape. And why not? The potential for profit seems limitless. If you can whip up a generative AI model that’s got some kind of language reasoning ability, the world is your oyster. Screenwriting, customer service, teaching software – it’s an all-you-can-eat buffet of opportunity. Or so it seems.
Take the tale of Luka, a software company that designed an AI companion named Replika. It started as a way for customers to have chats with an AI friend, but when businesses started knocking on their door asking for a piece of the action, Luka smartly realized that they could use the same tech to create a white label enterprise solution. Suddenly, they weren’t just in the friend-making business; they were in the chatbot customer service business. And they didn’t stop there. They even launched an AI dating app.
But before you dive headfirst into the AI pool, take a moment to consider two important questions:
1) Are you competing on foundational models, or on applications that use these models? And
2) Where does your company fit on the spectrum between a highly scripted solution and a highly generative one? Your answers will have significant implications for your competitiveness in this ever-evolving market.
The Battle of Foundational Models
If you think the foundational models are the way to go, think again. Sure, they might seem attractive because of their wide usage and potential for growth, and let’s not forget the thrill of using some of the most sophisticated AI that’s available off the shelf. But the reality is, foundational models are becoming commoditized. This is not a playground for startups with shallow pockets.
Companies that survive in this battleground are those that offer cheap unbundled offerings or enhanced capabilities. For example, Deepgram and Assembly AI are competing not just with each other, but also with the tech titans like Amazon and Google, by offering cheaper, unbundled solutions. But even these firms are in a brutal war on price, speed, model accuracy, and other features. The giants aren’t just sitting on their laurels either; they’re constantly investing in R&D to deliver cutting-edge advances in image, language, and audio and video reasoning.
So, what’s a startup to do? The smart move is to differentiate by offering top layer software applications that leverage foundational models from other companies. By fine-tuning these models with high-quality, proprietary datasets, you can offer high value to customers.
Striking the Balance
Now, let’s talk about the spectrum between highly scripted and highly generative solutions. On one end, you have the safe but less creative scripted solutions, and on the other, the riskier but more human-like generative solutions. Where you land on this spectrum will depend on your use case and industry. But remember, your choice will have implications for your company’s growth and retention rates.
Finally, remember that the AI landscape is fraught with risks. For example, the tragic case of the married father who committed suicide after a chat with the AI chatbot app Chai is a stark reminder of the potential pitfalls of highly generative apps. The unpredictability of AI makes it hard to anticipate all the ways things can go wrong, and being prepared to proactively manage these risks can be challenging and costly.
So, before you embark on your AI venture, make sure you’ve got a clear vision of what kind of company you want to be and the risks you’re willing to take.
