What has been achieved this year?
In our second survey results review before the year’s end, we consider Apryse’s survey of 465 organizations across North America, Europe, Australia and New Zealand. The global survey focuses on AI adoption and document infrastructure; however, our interest lies in scalability issues surrounding data quality.
But first, some background.
At a 48.4% representation, the audience was dominated by IT leadership, which isn’t really our purview but nonetheless roles covering product leadership (VP/Director of Product) and digital transformation/innovation lead, represented 14.2% and 8.4%, respectively, making the results of the survey relevant to us.
With regard to the size of the businesses, over 62% of respondents came from organizations with more than 200 employees, with 37% representing enterprises exceeding 5,000 employees, demonstrating that AI implementation is focused on companies with the resources and operational complexity that demand enterprise-grade solutions.
Who was surveyed?
Product leadership and digital transformation feedback is the most relevant to us.
Where are the participants?
North America, Europe & Oceania
- North America, 41.5% of participants, comprising the US and Canada showed the highest rate of AI production, which the report said demonstrated a more aggressive approach to commercializing AI applications.
- Europe, 38.3% of participants, (UK, Germany, France, Spain and Italy) was more focused on pilot and testing.
- Oceania, 20.2%, comprising (Australia, New Zealand) was more concerned about data quality and policy/security.
.
The crux of the matter
For us, the main insight the data brings is the scalability challenge, which comes a close second after privacy and security concerns (see image). With 76.6% of organizations reporting that between 25% and 75% of their total data is in documents such as PDFs scans and forms (see second image), organizations are investing in pre-processing tools—tools that convert these formats ‘meant for human eyes,’ into those that are machine readable, which means going beyond text scraping to include those that interpret layout and context.
The scalability challenge
Data quality
How are documents stored?
PDFs take the lion's share
This means that businesses require pre-processing tools, tools that interpret layout and context and beyond tools that merely scrape text. Organizations are investing in solutions and, according to the report, favor document structure recognition and document classification tools (see chart below). This shows that the focus is no longer about making text selectable but understanding where it resides (table, header, key-value field) to add semantic context for the AI.
Solving the problem
Which pre-processing tools are preferred?