How Search Engines Taught Me to Build AI Systems That Work
I started out in abstract mathematics, long before there was an Internet. Yet I quickly found my niche as soon as it existed: building search engines. I later sold my technology to Google, but that’s another story to be told another time. Right now, I’d like to discuss AI—because the reason your AI strategy is likely failing is because you don’t understand how search works.
Why is it that when you search for “tickets to Paris” you don’t get articles about Paris, Texas? Or when you search for a new laptop you don’t get articles about Bible Study? This isn’t because of machine learning; it’s because of human design. The truth is, if you tell a machine to crawl the Internet for information, allowing it to read anything and everything, you’re going to end up with a colossal mess. That’s why entire careers have been built on crawling based on specific criteria, categorizing it into specific buckets, and ranking it based on intuitive assumptions. If you do that job incorrectly, a simple search for a quick dinner recipe can easily turn into spelunking down the wrong rabbit hole.
The problem with AI is that most of us are failing to train it because we don’t know how. Think of a Hollywood celebrity. Back in the day, gatekeepers could control who would become famous. (Some would argue they still do.) But now, thanks to social media, people have come out of the woodwork, acquiring fame for questionable skills. Still, despite their lack of talent, no one can deny their understanding of how new media works and how to take advantage of it. Without a similar understanding of AI, you’ll end up being the random individual in the middle of nowhere wondering why their antics only get a handful of clicks. It’s because they don’t understand how the system works.
With our clients at Stealth Dog Labs, we train AI to work only with specific information about a company. We use mindset psychology and abstract mathematics to create authentic research files, training it to only produce results based on a company’s specific A+ customers, their products, their messaging, their brand purpose, and their passion. The intersection of these highly specific files generates unbelievable accuracy and clarity for clients in a way that almost makes the AI hype believable. However, if you merely ask AI a question and expect it to crawl the entire Internet for an answer, you’ll either end up commoditized, more confused, or swearing off innovation altogether.
Now, part of this is due to industry leaders claiming AI can solve any and all problems if you simply “throw more compute at it.” But that has not been our experience. Instead, most of the work we do is offline, presenting AI with very niche materials before allowing it to transform that information into a specific response. If you choose to do it the other way, hoping to skip the hard work and land on an easy solution, you’re basically telling the machine to go read the entire world and come back with the answer to your ultimate questions. Sadly, there’s so much noise online, the machine will struggle to find the signal. And when that happens, you’ll get an answer like “42” for your brand purpose—and for us, that’s just not good enough.
Of course, AI is an amazing innovation; but in our opinion, it’s simply the next frontier of search. As of today, this machine still lacks the inherent discernment and intuition that exists within mankind. And that is precisely why AI has failed to live up to so many expectations when it comes to solving the core business problems of today. Industry leaders promised magic at the touch of a button, enabling you to fire your workforce and cut costs like never before. Instead, AI is inviting you to work harder—not as a machine, but as a thinking, intuitive human. And if we are to squeeze the full utility out of AI, we need to stop looking at it as a god and start thinking of it as a search engine—one that requires conscious organization in order to work effectively. It may not be the sexy answer, but it just might be an honest one.