Articles
Based on my experience as the first general manager of IBM Watson (2011-14) where we essentially created the Enterprise AI category, to my current investment in nine AI companies, I believe that 2018 will be the year of pragmatic AI adoption. While the self-driving cars and sentient robots will continue to grab headlines, the capital requirements are high and returns are far from guaranteed. Unlike consumers and many in the media, businesses will sharpen their focus on using AI as a practical enterprise tool to power business processes and solve complex problems in spaces such as truly personalized user engagement, compliance, trading, and underwriting. These business processes may not deliver exciting science-fiction fun, but they are already delivering exceptional ROI.
The Challenge with Horizontal AI and Data Science
1. Data Types and Vocabulary_
Every industry has unique data types, vocabulary and high value data sources. Your AI platform should be optimized to work with these and include domain knowledge graphs to speed time to value
2. Regulations and Compliance_
Each industry has unique compliance and regulatory requirements for AI explainability and performance. Your AI platform needs to be able to support these requirements out of the box.
3. Business Metrics_
Business and ROI metrics for assessing AI’s impact and value generation vary significantly by industry. Your AI platform should have industry specific analytics to help optimize AI performance.
4. Human-Machine Experience_
Each domain has its own unique human-machine micro-interactions and your AI platform needs accelerators that model these interactions. For example, the way a derivatives trader wishes to consume AI powered applications is quite different than a care provider working with diabetic patients. In many cases these micro-interactions are not visually exciting, but they are essential for building user trust in AI and driving adoption.
In summary, industry-specific challenges require industry-specific AI.
Thursday, February 1, 2018
Five Practical Reasons for Industry-Specific AI Software in 2018
Much has been written about Artificial Intelligence (AI) this year – as a source of both huge excitement and apprehension. CEOs and Boards are now asking_ What impact will AI have on my organization, and how do I go about applying to drive business results and prevent potential business model disruption?
Based on my experience as the first general manager of IBM Watson (2011-14) where we essentially created the Enterprise AI category, to my current investment in nine AI companies, I believe that 2018 will be the year of pragmatic AI adoption. While the self-driving cars and sentient robots will continue to grab headlines, the capital requirements are high and returns are far from guaranteed. Unlike consumers and many in the media, businesses will sharpen their focus on using AI as a practical enterprise tool to power business processes and solve complex problems in spaces such as truly personalized user engagement, compliance, trading, and underwriting. These business processes may not deliver exciting science-fiction fun, but they are already delivering exceptional ROI.
The Challenge with Horizontal AI and Data Science
To solve real business problems with AI, you need an AI technology platform and a partner with a fundamentally deep understanding of your industry and its technology requirements. Employing a horizontal data science platform or consuming a general-purpose API-based cognitive service will not adequately solve the countless, complex problems that too often lead to expensive, time-consuming, and failed AI science experiments.
Using industry-specific AI software platform speeds implementation and ultimately results in a faster return-on-investment. Here are five practical reasons I believe your AI technology platform needs to be industry-optimized:
1. Data Types and Vocabulary_
Every industry has unique data types, vocabulary and high value data sources. Your AI platform should be optimized to work with these and include domain knowledge graphs to speed time to value
2. Regulations and Compliance_
Each industry has unique compliance and regulatory requirements for AI explainability and performance. Your AI platform needs to be able to support these requirements out of the box.
3. Business Metrics_
Business and ROI metrics for assessing AI’s impact and value generation vary significantly by industry. Your AI platform should have industry specific analytics to help optimize AI performance.
4. Human-Machine Experience_
Each domain has its own unique human-machine micro-interactions and your AI platform needs accelerators that model these interactions. For example, the way a derivatives trader wishes to consume AI powered applications is quite different than a care provider working with diabetic patients. In many cases these micro-interactions are not visually exciting, but they are essential for building user trust in AI and driving adoption.
5. Learning Curve_
AI is a complex domain and using industry optimized platforms reduce the “soft cost” with lost productivity and morale that comes with implementing a complex new system.
AI is a complex domain and using industry optimized platforms reduce the “soft cost” with lost productivity and morale that comes with implementing a complex new system.
In summary, industry-specific challenges require industry-specific AI.