How to profit from generative AI without the hype

Every technology has a hype cycle and it goes something like this:?

1) A technological breakthrough or product launch gets people talking about a new technology?

2) Expectations become inflated, sparking a flood of capital into start-ups?

3) A period of disillusionment grips early adopters and markets?

4) Productive use cases start to emerge

5) Mainstream adoption?

Applying this to generative AI, we’re probably straddling two and three. Stage one happened in November 2022 with the consumer launch of ChatGPT, which ushered in a wave of excitement—and panic—about potential applications.

The stratospheric valuations being applied to some AI start-ups suggest we’ve entered stage two. ChatGPT developer OpenAI’s valuation has reportedly tripled in the past six months to more than $80 billion, while 13 generative AI companies have reached a valuation of more than $1 billion, the vast majority passing the unicorn threshold within the last year.

But there’s also evidence we may be entering stage three. ChatGPT usage slipped for three months in a row through August, and the amount of time visitors spend on the site has been declining monthly since March. Others have pointed to the high costs associated with training and deploying the large language models (LLMs) that power generative AI.

Technology consultancy Gartner said in August that generative AI is in the Peak of Inflated Expectations and is 5-10 years from what it calls the Plateau of Productivity during which mainstream adoption occurs.

Many gen AI start-ups will over-promise and under-deliver as their dreams collide with reality. Others will succeed by maintaining a laser focus on creating ROI for their clients with practical features that solve pressing problems.?

Only by understanding the shortcomings and risks of AI as well as its incredible potential can we deliver safe, predictable solutions that enterprises can trust.

While public LLMs, such as ChatGPT, can converse on topics from particle physics to K-Pop, you wouldn’t want to lean on one for business-critical communication with your clients.

Private LLMs are trained on smaller, more specialist datasets that mean they’re less good at dinner party conversations but have a deep familiarity with their sector. DialpadGPT, our own private LLM that powers our whole range of solutions, knows all about business conversations, having been trained on more than 5 billion of them.

Dialpad uses AI to assist contact center and sales agents with real-time coaching, sentiment analysis and call summaries. The proven results are faster agent training, more productive calls, increased sales, and happier customers. All with plug-and-play simplicity and cost effectiveness. No hype required.

For many, the AI hype cycle will be something they experience first-hand, by investing in unproven, complex solutions that require huge capital investment, facing disappointment and disillusionment with the results and eventually finding their way to a more productive outcome.

For others, AI will be something that is embedded in their business in initially modest ways, but with real results that allow them to keep investing steadily without the write-offs that can too often result from overly-ambitious technology bets.

Which road are you planning to take?