5 things we learned about AI this year

July 31, 2019
A look back at what we’ve learned about AI and business in the last 12 months. 

Over the past year, MIT Technology Review Insights has conducted numerous studies on AI, the rate of corporate adoption, the leading use cases, and the impact on customers and staff. So what did we learn?

1)   63% of businesses in a worldwide survey to the MIT Technology Review Global Panel have a centralized AI plan or are using AI on a case by case basis. The most common areas of application are product or service development (33%), research and development (20%), and customer service (10%).

Which statement best describes your company’s strategy for the deployment of AI?

We have not yet determined whether AI could have value
AI will not have value for our company
AI will have value but we have not yet adopted AI
We are investing on a case-by-case basis
AI is transformative; we have a centralized roll out plan
We have incorporated AI in our operations
NOTES: % of respondents.

In what area is your company using AI the most?

Customer service
Supply chain
Sales and marketing
Product or service development
Finance
R&D
We are not using AI
NOTES: % of respondents.

Global Panel AI poll: full results

2)   Companies are making big AI-driven gains in efficiency and customer satisfaction. In a global survey of marketing and customer experience leaders, we learned that AI is a key tool in driving service at scale. Some 46% had seen significant improvements in customer call efficiency, followed by customer satisfaction and assisting customers migrate across channels (both at 36%).

Areas where AI investments have significantly improved CX processes and contact center operations.

Speed of complaint resolution
Brand recognition and visibility
Customer call efficiency
Overall agent satisfaction with CX processes
Overall customer satisfaction with CX processes
Agent & Employee guidance
Assisting customers migrating across channels
NOTES: % of respondents.

 Download Humans + bots

3)   A dearth of tech expertise and low-quality data are slowing implementations. The number one difficulty for AI adopters is the shortage of AI talent. Some 60% of global survey respondents say lack of experience is a key challenge. Insufficient data quality is the second greatest challenge for businesses (48%), followed by the high level of required investment (35%.)

What are the main challenges as your company deploys AI? (top 3)

Regulatory/data protection
Unions or workforce issues
Shortage of internal talent for AI
Slow decision-making at the exec level
Insufficient quality of data
High level of investment required
Internal resistance to change
Difficulty in making the business case
Other
NOTES: % of respondents.

4)   Business leaders are mostly unconcerned about the impact of AI on jobs in their organizations. The vast majority (60%) see overall headcounts stable or increasing. That said, nearly 40% note that a small number of roles are being automated. More common, is that jobs are being supported and made more efficient by employees having AI tools available.

In the last 12 months, have you seen any job roles lost to technology/automation in your organization?

No
Yes <5%
Yes 5 to 10%
Yes > 10%
NOTES: % of respondents.

In the last 12 months, have you seen job roles changing and being augmented by technology in your organization?

No
Yes <5%
Yes 5 to 10%
Yes > 10%
NOTES: % of respondents.

Global Panel FOW poll: full results

5)    Ethical concerns are featuring prominently in the minds of business executives. With AI technology developing at full tilt, there are increasing worries about bias, data protection, cyber-risk and the possible impact on society. In a survey of Asia-based executives, some 42% said what vigorous debate about AI ethics was taking place. The key ethical questions, according to Genevieve Bell, distinguished professor of engineering and computer science at the Australian National University who took part in an MIT Technology Review Insights research study, should be around the following themes:

  • Autonomy: Will AI systems be autonomous, and should they be?
  • Agency: Who will set limits and controls and ensure they are consistently applied?
  • Assurance: How will humans accommodate safety, risk, liability, trust, privacy, and ethics?
  • Measurement: How will we measure the effectiveness of AI systems?
  • Humanity: How will humans interact with AI-driven systems?

 Download Asia’s AI agenda: The ethics of AI

Download Getting smart about the future of AI