The Rollercoaster Ride of AI: Hype, Hope, and Reality

Society’s adoption of innovation seems destined to traverse a bumpy course from hype to disillusionment, yet beneath each cycle substantive progress accrues.

Change is messy. When companies roll out major new initiatives like reorganizations, IT system changes, or process redesigns, employees get swept away on an emotional rollercoaster ride. In the 1970s, psychologists Don Kelley and Daryl Conner coined this bumpy track, the Emotional Cycle of Change. This cycle applies not just to individual employees but also to companies undergoing significant change initiatives.

The cycle starts with the Honeymoon Phase, also known as Uninformed Optimism. Folks are amped about the change and have sky-high, often unrealistic expectations. But that excitement soon nosedives into the Hostility Phase, or Informed Pessimism. Reality rears its ugly head, and people realize the change is way harder than expected. Frustration sets in, morale plummets, and managers better watch out for pitchforks.

After a period of gritted teeth and complaining by the water cooler, employees enter the Integration Phase, known as Informed Optimism. People start accepting the change, learning new processes, and figuring out how to adapt. This leads to the Final Acceptance Phase, where the change becomes standard operating procedure. Employees regain their mojo and embrace the change.

If you've lived through a new system rollout or department reshuffle, you've likely cursed through your own Emotional Cycle. New workflows spark hope before drowning you in disruption. But eventually, you accept it and maybe even tell the newbie "that's just how things work around here now."

The Gartner Hype Cycle

Organizations go through similar tumult when adopting new technologies. In the ‘90s, Gartner’s Jackie Fenn mapped this journey in the Hype Cycle model. Emerging innovations trigger inflated expectations that inevitably slide into disillusionment before hitting mainstream adoption.

Like employees, technologies enjoy a honeymoon hype phase. But obstacles soon derail the party, spurring frustration. For every iPhone that survives the trough of disillusionment, technologies like 3D TVs and Google Glass sink into oblivion. With time and learning, the tech matures and integrates into standard workflows. Mobile phones and tap-to-pay made it up the slope of enlightenment.

It’s no surprise that generative AI landed squarely on the Peak of Inflated Expectations in Gartner’s 2023 Hype Cycle. AI is everywhere these days. It’s becoming like mobile apps - everyone wants it, even if they aren’t sure why.

Just look at the QITM XE PRO massage chair for sale at Relax the Back. It claims to use "AI technology" to customize massages based on your tension and stress levels. AI-powered furniture may seem frivolous, but it speaks to the hype swirling around artificial intelligence.

There is good reason for skepticism amid AI's oversold expectations, however. While generative AI demos seem impressive, many pitfalls remain translating flashy prototypes into production systems. Chief among them is large language models' tendency to confidently hallucinate false information, making business use treacherous.

Moreover, this is not AI's first time facing the peak of inflated expectations. The field's first "winter" occurred between 1974-1980, when early AI research showed promise but computers lacked processing power for complex capabilities. As funding subsequently dried up, once-promising projects shuttered.

AI went through a second winter between 1987-1993 with expert systems. Despite renewed excitement, these too failed to deliver and disillusionment set in.

The Substance Beyond the Hype

Amid the hyperbolic discussion of AGI and existential risk, something substantive however is unfolding - the democratization of AI.

In a July 2023 Stanford talk, AI pioneer Andrew Ng explained this change. Ng co-founded the Google Brain project, built the machine learning team at Baidu, and created the famous Coursera course on machine learning. He has witnessed the evolution of AI firsthand.

Ng explained how before Gen AI, using AI to solve business problems required expensive, months-long supervised learning. Companies had to manually label training data to build classifiers. This worked for high-value problems like ad targeting and social feeds impacting millions. But it was infeasible for smaller-scale issues.

Now, foundational models enable prompt-based learning that is accessible even to non-technical users. Per Ng, this allows affordable AI solutions for previously unviable problems.

For example, AngelList, a platform connecting startups and investors, deprecated its hand-engineered model for news classification. As described by AngelList's Thibaut Labarre:

"I spent years building News Article Classification models. Then, we were able to deprecate the whole thing and rewrite the whole system in a day. And I'm not kidding. Like, in one day. All by leveraging large language models like OpenAI."

While inflated expectations will inevitably lead to some failed AI projects in the trough, substantive retooling is underway.

Navigating the Hype Cycles of Change

Roy Amara was an American futurist who coined Amara’s law on technology adoption:

“We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.”

Amara’s insight rings true as we ride each wave of innovation. In the late 90s, e-commerce hype dwarfed reality. Pets.com and other failures plunged us into a dot-com winter. Yet, from the ashes rose companies like Instacart and Chewy.

This cycle continues today with artificial intelligence entering the Peak of Inflated Expectations. The Gartner Hype Cycle teaches us AI will soon fall into a Trough of Disillusionment as projects fail to deliver. But Amara’s law says not to lose sight that substantive change brews under the surface hype.