Companies are using AI to hit business goals, even though they can’t explain how it works

A new survey of tech decision makers finds that security concerns are high and trust is low when it comes to artificial intelligence.

AI (Artificial Intelligence) Concept Machine Learning, Nanotechnologies, Smart Objects and Global Network Technology

Image: KOHb, Getty Images/iStockphoto

Artificial intelligence (AI) algorithms are hard at work for many businesses, despite the fact that no one has figured out explainability, according to a new survey. Research from Modzy found that security is also a serious concern with 72% of companies reporting that security breaches or threats to AI systems. Modzy surveyed 821 decision-makers about how companies are deploying and managing AI for the new report, “The Race Towards Artificial Intelligence Adoption.” 

The survey found that 85% of decision-makers consider the explainability of AI to be extremely or very important. At the same time, leaders are also skeptical about the ability of organizations or government agencies to mandate this capability, with 59% agreeing it is impossible to do so.

The same percentage of respondents said that even data scientists do not fully understand how some algorithms work. Trust is also an issue for AI implementations with 64% of respondents saying employees in their organization still do not trust or understand AI-enabled recommendations. 

SEE: Natural language processing: A cheat sheet (TechRepublic)

The survey took place during October and November 2020 and included technology decision-makers at companies with more than 1,000 employees.

Despite these concerns, these AI techniques are in the widest use, according to the survey: 

  • Machine learning                          88%
  • Computer vision                            81%
  • Natural language processing        81%
  • Deep learning/neural networks     80%
  • Robotics and motion                     77%
  • Generative adversarial networks    66%

The survey found that these are the most common uses for AI technologies:

  • Analyze data/identify underlying patterns     60%
  • Improve planning and optimization               52%
  • Improve knowledge capture and extraction  48%
  • Engage with customers                                 48%
  • Automate back office processes                   44%
  • Analyze images or videos                              44%

Survey respondents were organized into two categories: Significant impact organizations (SIOs) and low impact organizations (LIOs). This reflects the importance of AI on a business’s operations. SIOs were much more likely to use AI for all of the tasks in the list above.

SIOs also were much more likely to see value from AI deployments with 75% stating they see significant value from the technology compared to only 30% of LIOs seeing the same value. There is also a significant difference in spending plans between the two groups with 88% of SIOs planning to increase AI investments over the next two years compared to 52% of LIOs.

Josh Sullivan, head of Modzy, said in a press release that successful adoption begins with a fundamental shift in thinking to overcome complexity obstacles.

“2021 will be the year when those implementing AI will start achieving value at scale, while those spending months training brittle models and failing to catch up will be at an increasing, exponential disadvantage,” said Sullivan.

The survey authors concluded that “AI technologies will only live up to their potential if organizations can develop greater explainability, trust, and security.” Companies that accomplish these goals as well as building employee and public trust in the technology will win the race to succeed with AI.

Also see