I'm coming to the conclusion that the expected wave of next-generation AI spending in both technology and technology services may last longer than both investors and companies are planning. In a recent blog, I talked about his realization that most (probably over 90%) of proof-of-concept (POC) efforts launched during 2023 will not be operational in 2024. We're spending hundreds of millions of dollars building AI tools and the capabilities to run them. What has changed? And where are the opportunities still to take advantage of this important technology?
Change in decision makers
We at Everest Group, in partnership with Yates Ltd. and CalypsoAI, convened a panel of 50 CIOs and CTOs from Fortune 500 and other large companies to explore AI initiatives and invest wisely in this rapidly changing technology. We continued to discuss ways to utilize this information.
It is clear that over the past few months, new stakeholders have come into the picture to decide who authorizes and funds Gen AI. And the questions they ask have also changed.
Additional decision-making authority is given to the department head. I'm not talking about the head of the business line. I'm talking about managers and departments, the budget holders in those departments.
It's very practical. Instead of focusing on where AI works, they want to know where AI works. People ask me questions like, “Where do other companies use people like me?” “How exactly are they using it in my set of responsibilities?”
If they believe that other people with similar responsibilities at large companies are using gen AI for specific tasks, they may want to fund or sponsor its implementation for those tasks. I'm very willing to do that. They are not interested in further experimentation with what the technology can do.
Additionally, return on investment (ROI) doesn't seem to be calculated much when making decisions. The ROI seems to be estimated. (“If others like me are using it for these tasks, I would seriously consider adopting it too.”)
Pricing and usage examples
Gen AI's current price is very expensive. This price seems to be starting to fall. I believe this efficiency will decrease even further, as we will soon become much more efficient in the use of computational power to run next-generation AI engines, rather than computing power. This makes your use case much more efficient.
Given that generation AI can potentially be 10x more efficient in using compute power, with only a small number of use cases being adopted, the next generation will create a situation of significant overcapacity. There's a good chance. 6-9 months.
This extra capacity could be a good thing for consumers, as it could lead to a significant reduction in the cost of using generative AI. But this could be a thorn in the side, or potentially a problem, for both high-tech and high-tech services companies that are counting on the next wave of AI to fuel their next growth.
Current pricing for most generation AI tools is too complex for department managers to understand. Pricing should be simplified to something much easier to understand and predict than the usage and token structures often found in Gen AI tools.
Pricing needs to not only be simple, but also low. I think prices are likely to fall in the short to medium term.
Impact of changes in decision rights
As explained above, decision-making power for funding gen AI is extended to departmental managers, and those managers are less likely to think about where AI will be used, but rather I started thinking about Ruka. They need to be clear about exactly which tasks in the chain of responsibilities it applies to.
This shift in consumer sentiment has very important implications not only for technology companies looking to sell products, but also for technology services companies looking to profit from the implementation and operation of these products.
The first implication is the need for a change in messaging. Currently, messages about the AI generation are focused on being imaginative about what AI can do. The announcement of ChatGPT being widely or freely available sparked that attention and caused industries and individuals to think more broadly about where it could be used.
Department managers don't want to go through this experience again. They are looking for practical answers that can be used today. Therefore, messaging needs to be brought down to a level that indicates exactly what it does, what tasks it enhances or removes, and which responsibilities are affected. Your message should also include assurance that other companies and people like you are succeeding.
The problem with this message is that there are countless departments with specific tasks that can and do apply artificial AI. Therefore, messaging is too complex.
Another factor is that implementing genetic AI is increasingly becoming a business problem rather than a technical one. Yes, there are some technical and security issues involved. But most of the issues are around integrating Gen AI into new works.
This becomes both a messaging and channel nightmare for existing technical messages and channels, as this requires not only technical experience and technical skills, but also domain industry knowledge and domain experience.
One possible answer is to leverage technology services partners to convey that message in a tangible way at the departmental level. Use these tools as channels to easily and tactically communicate with department managers regarding specific tasks that the tools may impact. You will also assist department managers with implementation, education, and change management tasks that increase business and consulting needs.
Therefore, we will likely need to rely even more on technology services companies to bring these tools to market.
Gen AI lacks the benefits that many historical technology implementations of technologies such as ERP and CRM offered. In many cases, spending on technology services increases by four times the cost of technology. However, that doesn't seem to happen with Gen AI as the implementation and SI requirements are not as great.
That said, the door appears to be open for multidisciplinary teams and consulting needs to help companies understand and begin using the tools. Therefore, the nature of the opportunity for technology services companies appears to be very different from that experienced with ERP, CRM, or other similar technologies.
Where do the opportunities lie in 2024?
Gen AI is a very important technology that will bring tremendous benefits to individuals and businesses. And some of that benefit will start showing up for him in 2024.
Currently, the most common use by enterprises appears to be enhancements or product enhancements to existing platforms. For example, Gen AI tools that improve sales operations platforms with all your sales data seem to be gaining traction in the marketing stack as well, and are gaining some traction.
This pattern applies when a company already has a technology stack with a large pool of data, on top of which AI generation tools can be laid to provide functionality and product enhancements to the existing set of technologies.
In other words, gen AI looks like a highly complementary technology to existing digital platforms. There's a lot of trust here.
An example is Microsoft Copilot. The data already exists on the Microsoft Office platform, and Microsoft Copilot is an extension of that platform. It adds additional value or additional usefulness. This tool improves the behavior of the platform.
There are still many such opportunities. Most companies have many opportunities to productively deploy generative AI, even with the high standards of only using tools that are already in operation by people at other companies in the same situation.
Looking outside of these situations, adoption cycles appear to be even longer, and it is unclear when large-scale adoption will occur elsewhere.