Employees attribute AI project failure to poor data quality.
A clear majority of employees (87%) peg data quality issues as the reason their organizations failed to successfully implement AI and machine learning.
For the Alation State of Data Culture report, Wakefield conducted a quantitative research study of 300 data and analytics leaders at enterprises with more than 2,500 employees in the U.S., U.K., Germany, Denmark, Sweden, and Norway. The enterprises were polled regarding their progress in establishing a culture of data-driven decision-making and the challenges they continue to face.
The old axiom garbage in, garbage out, is a key concern as enterprises ramp up their Artificial Intelligence (“AI”) initiatives. According to the report, 87% of respondents say data quality issues are a barrier to the successful implementation of AI in their organizations, with 46% saying they are very or extremely concerned. The report also found that just 8% of the data professionals surveyed say AI is being used across their organizations; 68% say AI is being used in some parts of the business.
Among other key findings in the report:
Inherent bias creates risk. 87% percent say that inherent biases in data being used in AI produce discriminatory results, creating risk for organizations. Solutions to this risk include:
Better modeling skills among analysts (42%).
Better curation and governance (38%).
Better literacy and understanding of data (38%).
Collecting data from more and more varied sources (36%).
Cataloging data for visibility (35%).
Core diversity in employees (35%).
Ability to crowdsource information (35%).
Stricter scrutiny of outcomes (33%).
Innovation and efficiency are primary drivers. When it comes to deploying AI, improving and innovating products and services is the top driver (43%), followed by improving operational efficiency (33%) and improving the customer experience (24%).
Skills are not the issue, executive buy-in is. 55% say getting buy-in from executives who control funding for AI is a bigger obstacle to using AI effectively than employees without skills to create AI models (45%).
Data quality issues are paramount. The top data quality issue to solve is inconsistent standards across data collection (50%), followed by compliance/privacy issues (48%) and lack of democratization or access to data (44%).
Success factors are many. Of organizations that have deployed AI, respondents cited better modeling skills among analysts (44%), cataloging data for visibility and access to available data (38%), and the ability to crowdsource info (38%) as ways to combat bias in AI. 31% say that incomplete data is a top data issue that leads to AI failing.
Sources: