Gen AI in 2024: Brilliant Outcomes or Merde Moments?

For evidence that generative AI is top of mind for enterprise decision-makers, look no further than Capgemini’s 2024 Investment Trends report, a summation of responses from 2000 business leaders across 15 nations and multiple verticals that was presented at the most recent World Economic Forum.

But there’s reason for pause when we look beneath the glossy numbers. (See below).

 The headlines:

• ·        Nearly nine-in-ten (88%) reported plans to increase investment in AI and generative AI.  No other IT investment topic received more attention. This is the sexy tech topic du jour; indeed, earlier Capgemini research showed that generative AI was on the boardroom agenda of 96% of surveyed organizations.

• ·        Nearly three-quarters of organizations (73%) plan to invest more in improving customer experience – a significant increase from the 20% who said so last year.   A previous Capgemini survey revealed that 71% of surveyed organizations said that generative AI would enable a more interactive and engaging experience for their customers.

• ·        More than eighty percent (83%) plan to increase digital investment this year, as compared to 39% in 2023.  Four vertical industries indicated interest above the norm:  consumer products manufacturing 88%), energy and utilities (88%), life sciences and healthcare (85%), and retail (85%).

Three thoughts in response:

• The survey results confirm what we have known: there is a tsunami of interest and exploration. But the survey results do not promise a corresponding flood of investment – only more investment than last year.
  
  Consider companies on a calendar fiscal; 2023 budgets were in review (and getting locked in) around the time of the November 2022 announcement of ChatGPT 3.5.   The original 2023 budget line for generative AI was probably $0. (In fact, there probably was not a line at all.)
  
  Yes, some 88% will increase investment in 2024. Keep in mind that even a little is more than none. We’ve now entered the enterprise trial and test period, the how-does-this-help-us-do-more with less phase.   

• The survey results point us toward a first use case of wide adoption – that of customer service and support. This is the realm of conversational AI, the hybrid of NLP-NLU-NLG and generative AI. It will be about chat and voice, of expanding conversational understanding (queries and in a range of languages), and it will be about ever-deeper integration into internal corporate data and retrieval-augmented generation from external sources. 

• The tsunami of interest and exploration – and a first focus on customer service and support – will bring fewer brilliant outcomes than desired, and many more merde moments than expected, as messy, dirty, duplicative, and unlabeled internal data soils and spoils initial results.
  
  I’m sure that few IT professionals are surprised by the fact that data cleansing and labeling was nowhere on the Capgemini list of priority increased investments.   
  
  Hardly a topic of bubbly board discussion. Yet absolutely essential to the achievement of replicable accuracy.

There are many, many steps to the Promised Land. We’re gonna have to take them all.

To be continued.

I’m Jon Stine, 35+ years in business and technology.

I read, I write, I advise.

Jcstine1995@gmail.com, +1 503 449 4628.

This article also published 2024.01.24 LinkedIn.