At the start of the previous year, who could have predicted that generative artificial intelligence and ChatGPT would take center stage?
A year ago, our projections hinted at a shift within data, analytics, and AI sectors towards simplifying and reimagining the modern data stack—an area of keen interest for us. The discourse around data mesh as a solution for data governance in distributed enterprises gained momentum, alongside the emergence of data lakehouses.
So, what can we anticipate for 2024? It comes as no surprise that generative AI will significantly impact database operations in the upcoming year, encompassing vector indexing, data exploration, governance, and database architecture. Before delving into the future, let’s reflect on how generative AI influenced our forecasts in the past year.
Insights into the Data Landscape in 2023
Referencing our previous year’s predictions, it’s evident that many of them materialized.
Progress was made in simplifying and streamlining the modern data stack through the extension of cloud data warehousing services, integrating transactions, data pipelines, and visualization from key players like SAP SE, Microsoft Corp., and Oracle Corp. Amazon Web Services Inc. notably enhanced its zero-ETL capabilities, bridging operational databases with Redshift and OpenSearch.
As anticipated, reality checks surfaced for data mesh as enterprises grappled with the complexities of implementing federated data governance. The concept of treating data as a product gained traction, although the definition of data products remained subjective.
Regarding data lakehouses, Apache Iceberg emerged as the standard open table format, bridging the gap between data warehouses and data lakes. Even Databricks Inc. made strides by aligning Delta tables with Iceberg standards.
A Shift in the Narrative
Initially, there was minimal mention of generative AI in the first quarter of the year. However, a significant shift occurred around April 1, as highlighted in our gen AI trip report. OpenAI’s ChatGPT, unleashed the previous November, swiftly amassed 100 million users in a matter of months, surpassing the growth rates of major social media platforms.
Subsequently, every data, analytics, and AI solutions provider scrambled to incorporate generative AI capabilities. Vector data support became a standard feature in operational databases, with English and other languages becoming prominent in application programming interfaces. The potential for generative AI to automate coding garnered substantial interest, despite concerns over intellectual property rights.
Generative transformer models extended beyond language tasks to encompass image creation, code generation, music composition, and data analysis. Amidst this, hardware, particularly GPUs, gained prominence, with Nvidia Corp.’s CEO making ubiquitous appearances at cloud conferences.
While the allure of Nvidia’s technology remains strong, the pursuit of alternative GPU sources intensified due to supply constraints. This scarcity led to enterprises committing to long-term GPU contracts, paving the way for potential aftermarket opportunities for unused GPU cycles.
The success of AI models, whether generative or traditional, hinges on data quality and relevance. In the realm of generative AI, the adage “garbage in, garbage out” remains pertinent.
AI venture funding trends 2012-2020 
Source: OECD.AI (2021), processed by JSI AI Lab, Slovenia, based on Preqin data of 4/23/2021, www.oecd.ai.
Envisioning 2024
A decade ago, data reigned supreme in venture funding, but AI has since overtaken as the focal point for investment. The OECD reports a significant surge in AI venture funding over the past decade, reflecting the sector’s rapid growth. Although recent years witnessed a slight downturn in funding, AI continues to attract substantial investments, with key players like OpenAI and Anthropic PBC securing substantial backing.
Looking ahead to 2024, we anticipate a proliferation of fit-for-purpose foundation models, marking a departure from the dominance of large models like GPT. This shift reflects a growing emphasis on optimizing training data for generative models, fostering a more balanced ecosystem between large and smaller models.
In the database domain, a trend towards stability is projected, with limited prospects for new entrants to disrupt the established players. The focus will likely shift towards AI integration within databases, paving the way for advancements in vector indexing and BI integration.
The Role of Vector Indexes and BI Integration
While vector indexes and gen AI-BI integration may not headline discussions, they are poised to drive significant innovation in the database landscape for 2024. Database vendors are expected to enhance vector index offerings and facilitate enriched gen AI queries with BI-style outcomes.
Vector indexes play a crucial role in optimizing similarity searches, with different indexing approaches catering to varying needs for recall rates, performance, and scale. As databases evolve to support vector storage, the differentiation in gen AI support will hinge on advanced indexing capabilities.
Moreover, the seamless integration of vector query results with tabular data will be a pivotal aspect of gen AI innovation in databases. This integration will enable complex queries that correlate summarized vector data with structured information, enhancing the overall database functionality.
Convergence of Data and AI Governance
A significant development anticipated in 2024 is the convergence of data governance and AI governance, bridging the gap between data administrators and AI developers. This alignment is crucial for tracking lineage and ensuring accountability across the data and AI lifecycle.
The integration of data and AI governance tools will facilitate comprehensive oversight, addressing issues of data quality, compliance, and model performance. By correlating data lineage with model training and deployment, organizations can enhance transparency and accountability in their AI initiatives.
Generative AI in Data Discovery and Governance
Generative AI is poised to revolutionize data discovery and governance processes by introducing natural language interfaces for query optimization and metadata management. Tools like Atlan showcase the potential for gen AI to automate data documentation and transformation tasks, streamlining DataOps operations.
The application of gen AI in database design is also on the horizon, with language models aiding in data structuring, schema generation, and synthetic data creation. While AI already plays a role in various database operations, the next wave of gen AI innovation will focus on content-related tasks, enhancing database development and deployment processes.
In conclusion, 2024 is set to witness a paradigm shift in the database landscape, with a strong emphasis on AI integration, vector indexing, and BI collaboration. The convergence of data and AI governance, coupled with advancements in generative AI, will drive transformative changes across the data ecosystem.
