One of the many amazing things about generative AI (GenAI) and the impact it’s having is how different it is from other recent big tech trends. Not only is it growing faster and more broadly than other buzzy technologies like blockchain and cryptocurrency, it’s also taking organizations into a great deal of uncharted territory.
First, companies seem to be ignoring several very legitimate concerns about what quickly adopting brand-new technology like GenAI may do.
A recent survey by TECHnalysis Research of 1,000 IT decision makers involved with their companies’ GenAI efforts (“A New Beginning: Generative AI in the Enterprise”) found that 99% of the companies who have started to use the technology still have issues with it.
In fact, three of the concerns they raised – data protection/security, inaccuracies, and copyright infringement – were selected by more than ½ of the companies using GenAI, as shown in Figure 1.
In addition, these same companies selected an average of 5 different concerns, reflecting the fact that the potential issues aren’t limited to a small set of possible problems but are more widespread.
Second, they’re also diving into GenAI headfirst without a clear understanding of how the technology works and what all its various deployment options entail.
To be fair, many technologies targeted at the enterprise haven’t been clearly understood, nor have their potential implications been fully comprehended at the time of their debut. With GenAI, however, the old rule about waiting to deploy something until it’s more mature seems to be going out the window.
There are several reasons why this is the case, but most of them boil down to the fact that GenAI has taken on an aura of inevitability and necessity that’s driving companies to start working with it sooner than they might otherwise.
The level of excitement around the technology – generally driven by some very impressive early experiences with it – has served as a particularly effective accelerant to adoption.
Indeed, the same survey showed that 95% of IT decision makers believe that GenAI could have either a profound or at least potentially noticeable impact on their business.
Not surprisingly, that’s led to a very palpable sense that everyone is adopting the technology (again, survey results showed that 88% of companies have started to).
It’s also led companies to believe that they could find themselves at a serious competitive disadvantage if they don’t move quickly – potential concerns and limited understanding be damned.
Of course, the tech industry is littered with new technologies that were initially expected to have a big impact on businesses. The difference with GenAI is that companies seem willing to overlook these potential problems because of the potential benefits it’s promised to unleash as well as the sense of urgency around the technology.
Another big issue with GenAI is the confusion around the subject. Though no one ever likes to admit that they don’t really understand an important new technology – especially people working in the tech industry – it’s clear that the confusion surrounding GenAI is still real and widespread.
Even something as seemingly basic as distinguishing the role and importance of a GenAI-focused foundation model (such as an LLM, or Large Language Model) and an application that uses the model can be easily misunderstood.
This too, is for several reasons. First, in the initial encounters with and discussions about GenAI tools like ChatGPT, these two elements have been used interchangeably.
It’s quite easy to think that ChatGPT is both the model and the chat-based interface that so many of us have interacted with. In truth, ChatGPT is the application, but it can be powered by various versions of the GPT large language model (such as GPT-3 or GPT-4).
This layered approach of an underlying engine and an application built on top of it is common to most GenAI applications. On one hand, it offers a new degree of flexibility, but also a much larger potential for confusion.
With typical enterprise software applications like CRM, office productivity suites, etc., for example, we’ve never had to think about the internal engine used to drive the functionality, nor have we had an option to change it.
With GenAI, however, you have the possibility of using several different engines for the same basic application, or the same engine across a radically different set of applications.
A single LLM, for example, can be used to create original text, summarize existing text, write software code and much more. As a result, the potential permutations can quickly become a bit overwhelming.
Another issue is that early offerings from different vendors are often described in relatively similar ways, but in fact may approach a problem or solve an issue in a completely orthogonal manner.
In other words, it’s often apples and oranges comparisons, which serve to further confuse the issue. This is one of the key reasons why educational efforts and clear, basic marketing messages are going to be so essential for the first few generations of GenAI tools.
Even people with decades of IT experience are discovering that GenAI is a different animal, and most need to have things explained in a simple, straightforward manner (whether they’re willing to admit it or not).
With the added pressure of moving quickly to make important strategic decisions around GenAI, the need for clarity is even more important.
Regardless of the concerns and educational shortfalls that may currently exist, it’s clear that the GenAI train isn’t slowing down in the least.
In fact, when the GenAI-enabled versions of the key productivity suites from Microsoft (MSFT) and Google (GOOG) (GOOGL) reach general availability – reported to be sometime this fall – I expect a new wave of GenAI buzz and activity.
These are the type of applications that organizations expect GenAI to have some of the biggest impact in – and they also happen to be the tools that virtually everyone in every business uses.
As a result, the real-world impact of what GenAI can achieve will never be more clear once a large number of people start using them on a regular basis.
The impact that I believe these tools could have won’t make the challenges or educational concerns discussed earlier go away – in fact, they’re likely going to become even more prominent.
Nevertheless, organizations need to prepare, and vendors need to focus in order to drive the kind of groundbreaking changes that GenAI is bound to create.
You can download the summary highlights of the Generative AI in the Enterprise study for free at this link: Generative AI in Enterprise Survey Highlights
Disclaimer: Some of the author’s clients are vendors in the tech industry.
Editor’s Note: The summary bullets for this article were chosen by Seeking Alpha editors.