信任第一：为人工智能驱动的企业成功培育数据治理文化

Trust is fundamental to successful AI. As digital becomes theprimary touchpoint for customers and partners, companies need Data is being rapidly democratized across the enterprise,spawning AI pilots and analytics use cases. But new freedoms can The data sources business users rely on can be siloed anddisconnected, with different practices, processes, and quality If trust in the underlying data isn’t rock solid, how sure can you beof any insights or outputs derived from it? Trustworthy data needs to be governed: managed, protected,compliance-checked, and quality-assured. Yet firms frequently Successful governance requires a culture shift and ongoingadaptation – paired with data management systems capable of Overcoming There are common barriers to successful data governance, someorganizational and some technical. On the organizational side,there are often cultural and behavioralhurdles to jump – particularly if one of the aims is to democratizedata. In the past, it was normal to see data as an asset to protect, notsomething to share. That can create a turf mentality between teams The common factor in both cases is trust. If people aren’t sure whetherthe data is well-defined, accurate, or suited to a specific use case, they’reless likely use it. If they can’t see the benefit of sharing their own data To modernize governance successfully, both the technology and theculture need to change. In both cases, senior leadership needs to be On the technical side,the biggest hurdle is typically a highly centralizedapproach to data governance. With the sheer volume and complexity oftoday’s data, centralized systems struggle to keep up. Legacy systems Data literacy asa foundation Many people lack the skills to: Understand which data is relevant Test the validity of the data they have Interpret data so the results from analytics are useful A/B test hypotheses to see which results pan out A core element of cultural change is data literacy. It’s more thanjust data awareness. Employees need to understand the who,where, what, why, and when of data governance so they can Or create easy-to-understand visualizations so leaders can Data literacy takes time. It requires upskilling – but also culturalchange. Organizations need to sustain an ongoing conversation Astudyby the US National Center for Education Statistics (NCES)measured the human ability to interpret data for decision-making Agentsneeddata Today, the industry is heading toward agentification of mostdata management tasks, an evolution aimed at achieving scale,efficiency, and continuous intelligence in how data is discovered,governed, and consumed. Without the ability to understand data The humanfactor:Coaching, When identifying internal data champions, thereare specific traits to look for: Areport from Capgeminifound that 75% of firms exhibitingdata mastery invest in building a collaborative data-first culture.Their aim is to make data a “habit” that’s engrained in day-to-day Look for people who are genuinely excited about drivingchange, who can influence the culture and help shift That’s why internal data and data governance champions are anessential part of any governance rollout. To embrace change,people have to live it first. Seeing someone you trust use new Look for people who are natural advocates, who commandrespect and who others naturally listen to. Look for people who are willing to lead by example. Use case-driven governance: Active data governance role-based coaching model When each role knows how to use the data governance platform, data moves from awareness to impact‘Role-play coaching’makes learning practical, engaging and actionable. As with any IT initiative, the rollout of a datagovernance modernization project should bedriven by specific use cases, whether that’s While data governance can support a widerange of objectives, it’s essential to clearlydefine your short-term and long-term goals.Metrics around effort and value must be Insurance industryspotlight:Enabling A US-based Fortune 500 property and casualty insurance companyhas set a strategic objective to empower data users and give them However, the change management program meant to unblockprocess barriers had stalled due to fragmented and inconsistent The Across the wider business, rules around data storage, stewardship,sharing, access, compliance, and usage had not seen full adoption.Many business units continue to run their own disparate datagovernance solutions, some of them built in-house. This created To address these challenges, the companyimplemented Informatica’s IDMC platform, ensuringproactive monitoring, automation, and consistent Without a consistent and proactive approach to governance, datawas often misinterpreted, leading to skewed decision-making, AI needs data, and data also needs AI. The success of AI projectsdepends on the availability of trusted and timely data used bydata scientists to train and scale their models. If data is missing