With the rise of artificial intelligence, an appreciation of the business value of data is now finally starting to increase. A global survey of 4,777 C-level executives and directors conducted by Accenture finds 90 percent of executives believe data is becoming a key competitive differentiator. A full 95 percent also said they believe generative artificial intelligence (AI) platforms such as ChatGPT will usher in a new era of enterprise intelligence.
The challenge is most organizations are not especially good at managing data. That’s an issue because the quality of the AI models organizations build and deploy are only as good as the data used to create them. One of the reasons there are so many reports of incorrect responses to ChatGPT prompts, for example, is that the model was built using a technique called Reinforcement Learning with Human Feedback. OpenAI, the research organization that created ChatGPT, essentially hired people to compare the results generated by two variations of their large language model as part of an effort to limit bias and toxicity, with mixed results. The content used, however, comes from sources of open data in addition to websites such as Reddit that were available at the end of 2021. As a result, ChatGPT doesn’t always distinguish between accurate data and propaganda, hate speech and marketing campaigns, and outright disinformation.
However, the large language model used to create ChatGPT is only going to be one of many. Advances in large language models are already making it possible to use less data to train generative AI models for specific use cases. There will soon be thousands of these models embedded within specific applications. In fact, open-source platforms for building large language models are already multiplying.
Data management skills will drive opportunity for MSPs
As intriguing as all these advancements are, the most important thing from a managed services provider (MSP) perspective is the impact they will have on the need for data management expertise. Data engineers are already one of the hottest job segments in all of IT, making it clear that demand is going to soon outstrip the available expertise. The opportunity for MSPs to plug that skills gap is going to drive new revenue streams for years to come. In much the same way fortunes were made by selling dungarees and shovels to miners during the gold rush, so too will MSPs with data management skills profit.
MSPs, if they have not already, would be well advised to start honing their data management skills. The data scientists that build AI models don’t want to spend their time managing data. The executives that hire them similarly don’t want to see some of the highest paid employees in their organization spending their time cleaning up data. Given the cost of hiring data scientists an MSP that can take on that task might just be worth their weight in gold.
In the meantime, it’s apparent that there is no application to one degree or another that won’t be augmented using AI models. When you consider the number of applications that are deployed, the data management opportunity for MSPs may soon prove to be nothing short of limitless.
Photo: VRVIRUS / Shutterstock
