CDOs are excited about generative AI (genAI), but a recent survey reveals a gap between enthusiasm and action. While 80% of surveyed CDOs believe genAI will transform their businesses, only 6% have deployed a genAI application. The culprit? Unprepared data.
Excitement Without Value
Generative AI thrives on unstructured data – text, images, and video. However, most companies haven’t curated their data for accuracy, completeness, and relevance to genAI models. This leads to poor-quality outputs. The survey reflects this challenge, with 46% of respondents citing “data quality” as the biggest barrier to genAI adoption.
Jeff McMillan, Chief Data Officer at Morgan Stanley Wealth Management, highlights the data quality efforts required:
“We’ve been curating document-based knowledge for five years… Every single piece of research content is reviewed… We had to spend a lot of time optimizing results…”
The DataOps Challenge
The survey highlights a disconnect between acknowledging data strategy’s importance for genAI (93% agreed) and actually making changes (only 37% agreed their organization has the right data foundation). Here’s where DataOps roles come in:
- Data Integration/Cleaning: 25% of organizations are tackling this crucial step.
- Data Exploration: 18% are surveying data for potential genAI applications.
- Data Curation: 17% are preparing documents and text for domain-specific genAI models.
Walid Mehanna, Group Chief Data and AI Officer at Merck, emphasizes the importance of data preparation:
“If we want to do AI, we need to build it on concrete, not quicksand… We’re working on data inventory, data fabric, data pipelines, and self-service insights generation.”
Prioritizing Data Domains
Given the data preparation workload, most companies should focus on specific areas where they expect near-term genAI implementation. The survey identified these as top priorities:
- Customer Operations (e.g., chatbots)
- Software Engineering/Code Generation
- Marketing & Sales (e.g., personalized campaigns)
- R&D/Product Design & Development
Should You Wait?
While we advocate for faster data preparation, we acknowledge other data projects’ importance (e.g., improving transaction data). Perhaps the slow pace reflects this, with 71% of CDOs prioritizing other initiatives for quicker value.
However, considering genAI’s transformative potential, waiting to prepare data isn’t ideal. The time to start is now, and getting a large organization’s data ready for AI could take years.
