Pitchbook predicts the market for generative AI in the enterprise will grow at a 32% CAGR to reach $98.1 billion by 2026.
I have been a tech entrepreneur for over 25 years. The pace of change in this space has always been incredibly fast. I used to tell folks I was operating in dog years, given that I would see about seven years’ worth of transformation in a single year.
The launch of ChatGPT late last year turbocharged that speed of innovation. Generative AI blew up, and every day major tech players like Microsoft, Google, and Salesforce released competing announcements of how they were integrating the tech into their platforms.
I have seen so much advancement, demand, and promise in generative AI since then, specifically on the interactive chat side, that I have started to measure the pace in hamster years, which is five times faster than dog years.
1. Attention will shift to training generative AI on enterprise data
Most of the tools that are making headlines work exclusively on data in the public domain. Yet there is a whole other world of possibility that opens as generative AI is trained on enterprise data. As Nicola Morini Bianzino, Ernst & Young’s CTO, puts it, this “will change the way we access and consume information inside the enterprise.”
This use case for generative AI is urgent because access to institutional knowledge is vanishing. Enterprise data is growing at an explosive rate, yet Gartner estimates that over 80% of that data is unstructured (i.e. PDFs, videos, slide decks, MP3 files etc.), which makes it difficult for employees to find and use.
Most information that teams create goes to waste because employees do not know what is available, or they simply cannot find what they need. Employees spend 20-30% of their workday tracking down information. When they cannot find what they are looking for, they disrupt the productivity of colleagues by asking questions or being directed to the resource.
Time is money, and as we inch closer to a recession, organizations are seeking new ways to drive efficiencies, lower costs and operate successfully with leaner teams. We will see more companies use generative AI to easily search for data within internal files and systems and empower the workforce.
2. Integration will be a key enterprise value driver
Today’s innovation is occurring within specific platforms. Take Microsoft, for instance, which is incorporating ChatGPT and generative AI into everything it offers. Recently, Microsoft announced Copilot 365, which can pull data from your Outlook calendar and emails to generate bullets for you to focus on in your next meeting. It can create Word and PowerPoint documents for you based on existing documents. These capabilities offer incredible value to users working within Microsoft’s tools. However, only 25% of enterprise data typically lives within Microsoft.
The rest of a company’s data lives in Google Drive, ServiceNow, SAP, Salesforce, Box, Tableau dashboards, third-party subscriptions and a wide variety of other systems. That’s why the enterprise value of generative AI grows exponentially when combined with federated search. It can pull data from a company’s entire set of tools and respond to a question or surface the information needed in the moment.
Think about how Roku brought streaming services together and made it easy for consumers to access all their applications in one place. That type of integration and innovation in generative AI will transform the enterprise.
3. Companies will start to establish generative AI strategies, policies and standards
This is the dawn of a new frontier for AI. Capabilities are now available that until recently were seen only in science fiction. Companies will need to understand the various use cases for generative AI and how this technology can increase productivity and drive growth. Organizations will need to establish policies on how to use the technology and will need to identify and adhere to the right compliance standards.
As companies adopt AI, teams leading the strategy and implementation will need to determine where it makes the most sense to augment existing applications, where to build new applications, and where to invest in packaged applications.
4. Accuracy will rule
Some organizations are hesitant to get on board with generative AI because it occasionally makes up answers. This phenomenon is known as “hallucination,” and it happens if there is not enough content available upon which to base a response or when the system believes that inappropriate data is the right data.
The challenge is that generative AI can confidently assert wrong or outdated answers as fact. The ability to provide evidence for answers will quickly become table stakes for providers of generative AI tools. Seeing exactly where the answer comes from enables users to validate the response before they act or make a decision based on inaccurate information. They can also tell the system if the answer is inaccurate, so the AI learns for next time.