From its start with efficient batch processing with data warehouses for descriptive analytics, and the inclusion of streaming data in real time to build recommendations, we find ourselves at the forefront of a new stage of evolution: generative AI (gen AI).
This generative powerhouse has fueled vertical integration, giving rise to industry-specific solutions that harness the full potential of generative capabilities and unlocked the imagination of many.
It started when one capable model suited for text gained mainstream attention, and now, less than 18 months later, there is a long list of commercial and open-source gen AI models are now available, alongside new multimodal models that also understand images and other unstructured data.
So how can organizations stay agile with change and innovation happening so quickly?
Recently Tom Stuermer, Global Senior Managing Director of Accenture, and Sridhar Ramaswamy, CEO of Snowflake, spoke on “Data Cloud Now” about the transformative nature of gen AI and large language models (LLMs). This blog highlights three key gen AI trends the two AI leaders see emerging.
AI is moving beyond the hype phase
For Stuermer, gen AI is having more than just a moment. It’s already having a real-world impact.
“It’s about the most transformative technology that I’ve ever seen in the period of time
that I’ve been around,” Stuermer said. “And yes there’s a lot of hype, but the demand [and] the
benefits we see from it, and the opportunities to utilize AI/ML and generative AI collectively inside of a data platform, are compelling.”
The numbers back up Stuermer’s belief. He’s had over 3,000 conversations with clients about gen AI that have turned into more than 700 projects using the technology.
Over the past years, Accenture has been working on gen AI projects that have been centered around proof of concept (PoCs) projects. Now in 2024, he sees clients asking, “How do I scale this?” He sees 2024 as the year of moving from PoCs to scaling to drive tangible impacts on diverse industries like manufacturing, retail and financial services. However, organizations need to prepare their data and workforce for more impact.
“There’s something we’re seeing…a whole other set of skills that we’re going to have to build,” Stuermer said. “As well as a lot of more opportunities that we’re going to have with our customers.”
It’s also important to slow down and analyze those PoCs to be realistic about what gen AI can do right now for organizations.
“Part of the difficulty with AI right now is figuring out where you can get the most business value,” Ramaswamy said. “It’s not like AI is going to instantly solve every business problem.”
Accenture and Snowflake plan to learn from the PoCs together to join forces on future gen AI solutions to create more value for their clients.
Elevating the capabilities of generative AI
LLMs are emerging as the new human-computer interface, bridging the gap between humans and technology using natural language. Organizations will find the most value from LLMs by using them first in areas like knowledge management applications, where a lot of information scattered across documents can now be accessible through simple-to-use chatbots. But people continue to be amazed by the breadth of gen AI’s capabilities, according to Ramaswamy, and he believes we’ll start seeing even more valuable AI applications coming to life this year.
“If you think about what we have done before, we typed on a keyboard and moved a mouse around to click on the screen,” he said. “Let’s face it, most people are not that
great at typing — it’s still a bit of a struggle.”
To him, gen AI and LLMs offer an opportunity for a better, more accurate, experience with machines. LLMs can be combined with complementary technologies, such as better voice recognition, to create a more fluid interaction with software.
“I think that means people are going to say a lot more,” he said. “I think the earliest examples we saw…is you ask [a chatbot] a question, ask it to write a piece of poetry. It’s like, here you go. You ask it a factual question [and] these models gave you an answer.”
Reimagining search and other experiences
It’s easy to get lost in headlines about the dangers of gen AI, but the technology is already transforming numerous aspects of our lives.
“Gen AI is real, and there are a number of no-regrets moves that you can take to prepare your data platform and start to actually incorporate gen AI into it so that you are better prepared to deliver gen AI for business value inside your organization,” Stuermer said.
In fact, he says there are 42 interventions Accenture has already identified some interventions its customers can perform today to better prepare their data platforms for gen AI. Ramaswamy added that there are several gen AI capabilities organizations can either quickly implement or modernize, like search and information retrieval.
“None of us needs to be in the mode of, I type something into a search box, I got
back 20 links [and] I need to now click on eight of them to figure out what it is that I want,” Ramaswamy said. “That day, you should know, has passed.”
Instead, organizations can give you a good retrieval system and an LLM, which Snowflake
provides natively, to simplify this problem. Still, organizations need to be realistic and implement realistic solutions while emphasizing value creation and managing expectations around gen AI.
“I think having that mental model [of] what’s doable and what is hard, what is fiction, I think having that perspective and building on it is going to help practitioners a lot, and the
executives that are going to be funding the practitioners,” Ramaswamy said.
Data’s role in driving the future of gen AI
Ramaswamy sees gen AI technology moving beyond the traditional question-and-answer format we see today. He believes models are poised for a tool-use revolution, with their ability to generate structured queries, such as SQL, and seamlessly interact with APIs. Gen AI will become the glue for interoperability, in his view, unlocking massive value across diverse enterprise domains.
“The classes of applications where I think we will work together on first and foremost are the
information applications,” he said. “[We’ll ask our models for] better answers in this specific area I’m looking for: help with a product that I have; look through all of your product documentation; look through all of your support cases and help find the most relevant things for me. Help me solve the problem that I have. It’s things like that that I think will have a big positive and relatively immediate impact with AI.”
These experiences will rely on data to power them. Specifically, Stuermer said many of the gen AI solutions that are popular today — like call center interactions and chatbots — have all created a demand for better data to be able to support the innovation Ramaswamy expects.
“There’s this reinforcing cycle that the more gen AI is getting deployed, the more demand there is for accurate, vetted, comprehensive data to prompt and train those models more effectively,” Stuermer said.
As demand increases for these tools, it will continue to put strain on the data platforms being used today. But Stuermer sees this as an opportunity and believes we’re already seeing data platforms, like Snowflake, increase their innovation to fuel the gen AI revolution.
As we move from gen AI hype to reality, organizations should focus on three key areas to successfully unlock its benefits in 2024, according to these two leading AI experts:
1. Organizations can reimagine the user experiences, enhancing it in particular where scattered information delays time to action, including search and information retrieval.
2. Organizations need to ensure their data is ready and that its high quality, to deliver accurate and reliable results for their gen AI solutions.
3. Organizations should adopt a data platform strategy that makes AI accessible to everyone, not just developers, thanks to the new natural language interface of models.
Watch the full Data Cloud Now interview to learn more. Ready to dive deeper into gen AI? Download The Essential Guide to Generative AI to learn how to prepare your organization for the next evolution of gen AI and how Accenture and Snowflake can help you get there. And learn more about the partnership between Accenture and Snowflake.
