All the major tech and financial leaders descended on Davos this week for the World Economic Forum. Energy, Trade, and Sustainability topped the official agenda, and I'll let other industry professionals comment on the implications for energy markets and renewables in those communities. For Digital Utility members, I thought it was enlightening that a quick search of WEF for "Artificial Intelligence" listed 46,300 results, with over 100 articles posted just this week during the event.
It's clear that AI has eaten the world and that includes most of the sessions at the WEF this week. As I listened to the speeches and pundits on the morning talk shows, I heard a common refrain. Cisco CEO, Chuck Robbins, best summarized the news. Robbins said that despite the hype, AI is nowhere as close to the use cases as the market might say it is. He commented that AI is still in its infancy, still defining the major enterprise use cases, still struggling with accuracy and quality, and still working to find its footing.
That's good news for utilities considering applications for the technology. No need to buy into the hype and rush into Artificial Intelligence. Major firms and industries have been investing and experimenting for a decade on generative AI, and the use cases are still being defined.
A few we hear about in the "hype cycle" that need more time and consideration for our industry:
- AI Will Replace Customer Service. Well, not yet at least. Gartner recently reported that only 25% of customers who used a chatbot on a customer service experience would opt to try one again. Some surveys report more than 70% of customers just don't like chatbots, and AI inaccuracies and hallucinations aren't making a great case for replacing traditional customer service with even the more advanced chatbot applications. Chat's certainly getting better, and LLMs trained with specific utility information like transcripts of customer calls with positive outcomes, utility policies and procedures, website content and program information can be incredible tools to help improve customer service team performance. AI will absolutely improve the performance of the customer service center long before it replaces a customer service representative.
- Utility Marketplaces Need AI to Make Product Recommendations. Not by a long shot. We saw a LOT of requirements in RFPs this year about AI capabilities for utility ecommerce sites, but the expectations and investment appear very premature. Amazon has invested in machine learning and generative AI tools to match customer profiles with suggestive merchandising on its ecommerce portals for at least a decade and only just launched an interactive AI for its shopping app, and then only in limited release. Amazon has incredibly detailed data on the purchase patterns of its customers, along with the browsing and viewing habits of Prime members, along with its own massive catalogs and augmented by 2x-5x as many products from third party sellers on its platform. Amazon feeds all of that data into its AI to manage the pattern matching and recommendation engine to try to improve conversion a few fractions of a percentage point. Utility marketplaces do not suffer any of those challenges. Catalogs on utility sites trend toward more manageable categories with a few dozen SKUs proven to appeal to utility customers. Utilities reluctant to integrate customer data with a trusted third party surely will balk at training a shopping AI with the intimate level of detail required to make it perform well on the marketplace. Utilities interested in better performance from their utility marketplace should first look at expanding their catalog to match customer needs and interests, then increase both the volume and personalization of their outreach campaigns to drive more awareness and customer traffic to the site. It's a much more cost effective and appropriate step to spark better performance. AI recommendation engines add a massive investment to an already costly revenue center for most programs.
- AI Revolutionizes Energy Efficiency. Well, maybe. CES demonstrated some terrific technology boasting AI driven advances in energy efficiency and energy management for C&I and building management that likely can deliver impressive benefits to new structures built with these new technologies. I am not sure that the cost of retrofitting aging structures generates sufficient savings to justify the investment at present energy costs. As the cost of the technology inevitably declines and the adoption rate increases, especially in business and industrial settings, it's likely that the pattern matching capabilities of machine learning and AI will have a net positive effect on energy management and efficiency. It's a struggle to expect AI to have much of an impact on home energy management and energy efficiency in the medium term except in the largest and most tech-enabled homes, and then only on a modest scale. I suspect that the AI features boasted about at CES make for more marketing buzz than added value, but time will tell. At least the demos were cool.
Utilities have an understandable and appropriately cautious approach to AI. Going into Davos, a review of the news and agenda would have us think AI was taking over the world. It was really great to hear Mr. Robbins more grounded opinions bring inflated expectations of AI back down to earth, at least a bit. Utilities have plenty of safe and affordable ways to experiment with the new technology where they can control the outcome and protect their customer and their infrastructure consistent with industry culture. How will you be using AI in 2024?