Key Takeaways from McKinsey’s Recent Report on AI

This is one of the better reports that I have come across (see the link below) and trust me I come across a fair number. It is a fairly long report containing a wealth of market information about the adoption of AI by various companies and several major segments. I find the information unbiased and a compilation of positives and negatives. Before getting into the nitty gritty of the content I would like to jump to the “Bottom Line”. 

The Bottom Line

Despite the tremendous hype behind AI, there is absolutely no question that the phenomena is real, it is big, and it is getting bigger with a ferocious pace. Take it seriously, embrace it or risk obsolescence. Reading through the plethora of use cases, I get the impression that there are three types of companies (when it comes to adopting AI). The first category consists of organizations that are clueless and clearly don’t really know what they don’t know. The second category is a collection of companies that are well aware of AI’s benefits but seem to be stunned and unsure of their next steps. Finally, there are two dozen companies that are extremely aware and are spending tons to acquire and/or build AI prowess. They are using AI in multitude of applications and tend to be serial deployers. What concerns me the most is the vacuum between the second and the third category. The gap between ‘haves’ and ‘have nots’ is extremely large and growing. In a way, I see a need for a (non-existent) forth category that should be squeezed between the former and the latter.

The following are just a few takeaways that caught my eye:

1. Investments made in AI in 2016 were in the range of $20B to $30B, 90% of which was in R/D and deployment and the remainder in M&A. VC and private equity investments were $6B – $9B

2. Only 20% of companies (out of 3000 surveyed) utilize AI in an operational capacity of some sort. Only 12% of documented use cases were in actual deployment

3. Sectors that have embraced digitization are far more likely to benefit from AI. High tech, financials, and telecom leading the pack

4. Companies with strong digital capabilities have been far more successful in their AI initiatives and typically commend higher profit margins among their peers

5. Tech giants such as Amazon, Apple, Alibaba, Baidu, and Facebook are dominating when it comes to both operational utilization and investment

6. Early adopters of AI tend to become serial adopters

7. The level of investment in AI has been mostly in companies addressing AI as a whole (developing technologies applicable to many areas). Vision, natural language, and autonomous driving come next

8. Most vibrant geographies: Beijing, Boston, London, New York, Shenzhen, Silicon Valley

9. Data to AI is like blood to the body. An organization lacking access to meaning operational data can’t benefit much from AI regardless of their AI prowess. Companies that are on the fence, should at least start accumulating relevant relational and structured data. This should explain the tremendous power that companies such as Google and Facebook posses

10. The big opportunity for AI upstarts is to understand the inner workings and intricacies of specific chosen industries and apply their AI technologies to solve real problems able to yield real tools real ROIs. Their technology is almost secondary

11. The concern for AI eliminating jobs is highly exaggerated and frankly mostly inaccurate. Only 1/5 of the use cases deployed by serious players are targeting labor efficiency

The following are some prominent use cases in Education, Energy, Manufacturing, Retail, 

Retail

> Forecasting customer trends

> Optimize logistics, and warehousing

>   Set adaptive pricing

>   Personalized and timely promotions

>  Predicting customer’s “next order”

>  And many more . . . . 

Electric Utilities

>   Accurate electric supply and demand prediction

>   Yield optimization, predictive outage, preventive maintenance

>   implementation of dynamic pricing

Manufacturing

>   Operating smarter, nimbler, and less prone to error. This can range from process automation, inventory management, and having robots that can adapt to their surroundings able to do far more than repetitive tasks

>   Reducing the number of iterations in the R/D process. One telling example comes from no other than intel. Heavy use of machine learning and data analytics enabled them to improve the yield of a certain process by 10%.

>  Untangling procurement flow to get better grip on the real cost of goods and services

Educations

>   Virtual tutors powered by AI can personalize learning and optimize teaching effectiveness

>   Educators are able to use personal, academic, and professional data to ensure the students benefit from choosing courses

>   There are already tools deployed in certain Universities that identify and engage with students at risk for quitting

>   The ultimate goal is to develop virtual teachers able to deliver the right material to the right students based on their pace and detect and correct problems before they become unsolvable.

Please follow and like us: