I have been calling my shots on generative AI's progression since ChatGPT's release last year. This article is the latest in a series that separates hype from reality. In February, I released a course on monetizing generative AI. Many reviewers and people who took it when it was first released have said similar things. The trends I explained are playing out today, and the course content is remarkably accurate.
With the latest update to ChatGPT Plus, OpenAI's focus on improving the model's ability to code and analyze data from spreadsheets has become public. I wrote this to a reporter on the 19th, "{GPT} will make…low or no-code solutions obsolete." Yesterday, many people who got early access to GPT's upgraded coding features said the same thing.
In this article, I'll explain a little bit about how I know what's coming. The short answer is, 'I've seen this before.' I will cover the impacts by focusing on products and what generative AI enables. In 6-months, the hype bubble will burst, and I'll explain the effects.
In the 12-month section, I cover specific use cases like hyper-personalization and a wave of personal assistants. The biggest reveal is that generative AI's capabilities have already peaked.
In the 24-month section, I explain what happens when generative AI matures and is widely adopted. Peaking capabilities sound like a bad thing, but in reality, it will be a massive tailwind for adoption and product development.
I am looking at offering another live course for 1) Data and AI Strategist Certification and 2) Data and AI Product Management. It’s your last chance to reserve your seat and early pricing. Email me (info@HighROIAI.com) and let me know what class, day, and time work best for your schedule. Subscribers get 15% off the early pricing but only for a limited time.
Learning From Experience
The hype is clouding the reality of generative AI's impacts. Don't get me wrong, I believe generative AI is a leap forward and will deliver disruptive products. It's not an Earth-shattering technology or once in an epoch disruption. Generative AI is the end of the beginning for AI and advanced models.
The beginning was IBM's Watson back in 2011. It appeared on Jeopardy and won against prior champions. That was the beginning of Chat GPT's story, and it faded away. IBM attempted to productize Watson across several use cases and domains but failed to gain significant traction. There was a lot of hype then too, but it resulted in little substance.
I saw my first AI hype cycle in the 90s. Microsoft began working on machine learning, and I learned about the field. I expected to join Microsoft upon graduating from college and build AI. The only product to come from that hype cycle was Clippy. My data science career would have to wait almost 15 years to become a reality.
The 90s revealed significant limitations that prevented machine learning from moving forward. Hardware, tools, and data needed to take a step forward before machine learning would become feasible. However, the interest that the cycle generated led to the advances of the early 2000s. It's why IBM began building Watson, and people like me were ready when machine learning became feasible in 2011.
Watson spawned a second wave of interest and advancement. It's one of the key reasons I got into the field and launched V Squared. It was the proof of concept I needed to see that convinced me products were now possible.
11 years later, generative AI has enabled a new generation of products. The theme of hype cycles is acceleration. Progress moves quickly, and the next generation of advancements leads to even more amazing results. It took almost 20 years to get from the 90s hype cycle to Watson and a decade from Watson to GPT. Follow the pattern, and we are probably 5 years away from the next leap forward.
Generative AI's hype cycle will be different because the technology is ready for prime time, and products will ship during this hype cycle. The next 2 years will be exciting. In this article, I'll introduce what will be built. The hyperbolic predictions will not come to pass, but big things have small beginnings. The modest products we see in the next 6 months will become massive disruptions over the next 2 years.
Six Months
According to an MIT and Stanford study (link to original publication behind a paywall), customer service agents using ChatGPT were 14% more productive than people who didn't use it. As adoption rises, a growing number of jobs will see similar productivity gains. Over the next six months, the productivity boost will spread to more roles and industries but not increase very much.
We're still early in the adoption cycle and may get up to as much as 25% more productive as people become more proficient with generative AI tools. That doesn't sound like much, but it's deceptive. Even a 15%-25% boost in worker productivity is a significant improvement for such a short amount of time.