3 Things We Learned at Oracle’s Data & AI Forum

On January 18, 2024, Oracle held its Data & AI Forum in Boston. Learn about our leadership’s main takeaways from the event, including the ways AI can boost efficiency across an organization, how to approach AI security, and the rise of retrieval-augmented generation in business AI.

2023 was, in many ways, the year of AI. But as large language models (LLMs) and generative AI technologies like ChatGPT captured the world’s attention, many businesses were left scratching their heads: How could these exciting tools be put to work in their organizations? 

Since last year’s explosion of AI hype, many of BSD’s customers have asked me similar questions. That’s why we were thrilled to be invited to Oracle’s Data & AI Forum in Boston on January 18, 2024. Attending this event provided us with insight into the current state of business AI—as well as where it might be headed. 

We’re excited to be sharing some of those insights with you here, including what Oracle’s experts had to say about where organizations can implement these tools, how they can boost efficiency, and which developments represent the future of business AI.

1. AI can boost efficiency across the organization

Throughout the event, speakers emphasized how organizations today might use AI to increase productivity—as well as just how groundbreaking these technologies really are. Much as Erik Bergenholtz, VP of Strategy and Operations at Oracle, would tell us: “While classic machine learning was focused on prediction, generative AI is focused on innovation.” 

Building on this, the day’s presentations highlighted how AI can be useful at every organizational level. Some of the examples that stuck out to me were:

AI for Sales

Sales teams can rely on AI to quickly identify potential leads, personalize pitches, and streamline communication. For example, these tools can suggest products or services that meet the needs and preferences of specific customers and generate proposals.

AI Chatbots for Employee Onboarding

Organizations can use AI chatbots to make employee onboarding more efficient. This can free HR employees from needing to process documents, suggest relevant training materials, introduce new hires to key contacts, or guide them through role-specific onboarding, as each of these tasks can be automated.

AI for Procurement

When it comes to procurement, AI tools are capable of analyzing company spending patterns, identifying opportunities for cost savings, and promoting more effective budget allocation. They can be used to find and evaluate suppliers as well as to support negotiation processes with data-driven insights about the market. AI-powered chatbots can also handle routine procurement inquiries from internal stakeholders and automate related tasks such as purchase order creation, invoice processing, and payment.

Across these solutions, we saw that AI use in business isn’t replacing humans—it’s enabling them to be more productive and innovative.

2. AI security remains a concern for businesses

Because many LLMs are trained on public data, however, they’ve been known to pass false information off as if it were true. As LLMs’ training data can vary greatly in terms of quality and veracity, they might inadvertently learn and reproduce inaccuracies or misinformation. This training data can also contain cultural and social biases, leading certain models to reproduce them with skewed or unbalanced information.

For businesses, public LLMs present additional risks. Given that these models often indiscriminately store and use users’ data, they can lead to the inadvertent disclosure of confidential, proprietary, or personally identifiable information (PII). 

In response to such issues, Oracle has designed enterprise solutions that minimize the privacy risks associated with AI use. 

One example is Oracle’s Cloud Infrastructure (OCI) generative AI service, which allows organizations to deploy private, pre-trained AI models that are tailored to their specific needs. These models can be trained on relevant industry-specific and organization-specific data. In addition to producing more accurate and useful outputs, this solution ensures that organizations’ data remains secure by granting them full control over its use and access.

3. RAG is revolutionizing enterprise AI 

Oracle’s OCI service makes use of a new approach to enterprise AI that nearly every speaker would comment on in some way: retrieval-augmented generation, or “RAG.” 

Whereas classic LLMs can’t fact-check in real-time or verify the truthfulness of the data that they’ve been trained on, the RAG approach optimizes AI output by retrieving information from external datasets or knowledge bases. This allows LLMs to access data that is current, relevant, and accurate, including recent market trends, internal reports, and up-to-date financial or customer information. In addition to providing businesses with trustworthy virtual assistants, RAG saves them money that would otherwise be spent on retraining models.

It’s no surprise, then, that major vendors like Oracle—as well as Microsoft, Google, and Amazon—have already begun implementing RAG into their enterprise AI solutions. For its part, Oracle offers support on every level 1) the pre-trained LLMs central to its OCI AI service; 2) AI Vector Search, which introduces semantic search capabilities to its proprietary Database 23c and is capable of augmenting RAG-powered LLMs; and 3) Fusion Cloud Applications, which use retrieval-augmented AI services to boost efficiency and improve decision-making processes.

If the Data & AI Forum was any indicator, RAG may be the way of the future for businesses hoping to streamline operations and gain a competitive edge with AI.