Altair Survey Reveals Opportunities To Improve Success Of Enterprise AI, Data Analytics Projects
Project successes suffer due to organisational, technological and financial friction.
Computational science and artificial intelligence company Altair Engineering Inc. has released results from an international survey, which revealed high rates of adoption and implementation of organisational data and AI strategies globally. The survey also showed that project successes suffer due to three main types of friction: organisational, technological and financial.
The independent survey of more than 2,000 professionals across ten countries, including India, and multiple industries showed a high failure rate of AI and data analytics projects (between 36% and 56%) where friction between departments exists.
“Businesses must make the shift to self-service data analytics tools that empower non-technical users to work easily and cost-effectively across complex technology systems and avoid the friction inhibiting them from moving forward,” said James R. Scapa, founder and chief executive officer, Altair.
Organisational Friction
The survey found that organisations are struggling to fill data science roles, which is a significant cause of friction.
75% of respondents said they struggle to find enough data science talent.
35% said AI literacy is low among majority of the workforce.
58% said shortage of talent and time taken to upskill employees is the most prevalent problem in AI strategy adoption.
Technological Friction
More than half of the respondents said their organisation often faces technical limitations, that are slowing down data and AI initiatives.
Respondents struggled most with data processing speed, making informed decisions quickly and experiencing data quality issues.
63% said their organisation tends to make working with AI-driven data tools more complicated than it needs to be.
33% cited legacy systems’ inability to develop advanced AI and machine learning initiatives as a recurring cause of friction.
Financial Friction
Despite organisations’ desire to scale their data and AI strategies, teams and individuals face financial obstacles.
25% cited financial constraints as a friction point that negatively affects AI initiatives.
28% said leadership is too focused on upfront costs to understand the benefits of investing in AI and machine learning.
33% said the “high cost of implementation” — real or perceived — is a shortfall when relying on AI tools to complete projects.
Optimism Despite Project Failure
Organisations across industries and geographies persist in using AI, despite high project failure rates.
One in four respondents said more than 50% of their projects fail.
42% admitted they experienced AI failure within the past two years; among those, average failure rate was 36%.
Despite failures, organisations continue AI use because they believe there is an opportunity to level up capabilities or services in the long run (78%) and its minor successes have shown potential for long-term breakthroughs (54%).
Many organisations struggle to complete their data science projects as well.
33% said more than 50% of their data science projects never made it to production in the last two years.
55% said over a third of their data science projects never made it to production in the past two years.
67% said more than a quarter of projects never made it to production.
Friction Exists Globally
Globally, the survey revealed that technology and talent are pain points when deploying organisational data and AI strategies.
Respondents in the Asia-Pacific and Europe-Middle East regions experienced more AI failures in the last two years (54% and 35%), compared to the North-South America region (29%).
65% of APAC respondents and 61% of EMEA respondents agreed their organisation makes working with AI tools more complicated than needed.
78% of APAC respondents and 75% of EMEA respondents said they struggle to find enough data science talent.
As per the survey, when organisations achieve “frictionless AI,” data analytics becomes easy, with projects that are quick, repeatable and scalable. There is no technical friction between them and their data, no organisational friction between data experts and domain experts, no workflow friction between data application design and production deployment and no migration friction when infrastructure or tools change.
