Wise Leadership & AI: Making Friends with the Machines

Only 28% of digital executives recently surveyed by Amrop agree their boards fully understand the meaning and scope of digital — let alone AI. Yet digital leaders know that the relationships between humans and AI have vast potential. Getting it right will mean better using the strengths of each and optimizing the synergies between them.

1 - We can see AI as a talent management task: putting the right ‘mind’ on the right job

Machine talent is best at rapidly identifying patterns in big data. Human talent, at identifying complex, causal relationships and imagining new avenues for value creation. Organizational leaders understand the need to streamline companies for human-to-human collaboration. The same applies to human-to- machine partnership. At the highest level, it’s also critical to align human and organizational objectives with the abilities of self-learning machines.

2 - Ecosystems are fertile territory for explorer organizations.

The gig economy with its dissolving borders is one example. Seven of the world’s largest companies (and many of the most profitable), are now platform businesses, large, scalable ecosystems such as Uber or Airbnb that connect suppliers and customers through their enormous networks. These provide access to exponentially more data than legacy organizations, with a diverse pool of resources, enabling rapid experimentation and expansion.

3 - The most promise lies in AI augmenting human abilities

It makes work more accurate, productive and enjoyable. One way of seeing intelligence is ‘doing the right thing to meet one (or more) objectives’. Artificially intelligent machines do this by supervised learning, applying statistical techniques to crunching (complex, big) data – finding patterns faster than humans can. It is in this way that they radically extend human ability, in voice or image recognition, for example.

4 - We must remember and manage AI’s darker side

The use of (historical) data may amplify biases such as gender or racial prejudice. And this can prolong unfair or unethical situations. Privacy concerns, data manipulation and cyber-security issues are well known risks. Furthermore AI- based tech firms. Google, Facebook, Microsoft, Baidu, Alibaba and Amazon are experimenting with incursions into our lives, manipulating and modifying how we feel, think and behave. In the quest to raise the probability of guaranteed outcomes and monetize them, these giants risk undermining our fundamental freedom.

5 - The effects of AI on society can swing both ways.

On the downside, and especially in the short term, AI risks employee inequality, with job security fears intensifying people’s risk aversion. On the upside, it has huge potential to enhance human well-being, leisure time and livespans, and improve the health of the wider environment.

6 - As in the last industrial revolution: there’ll be losers, but also winners

As Andrew Ng put it: “If a typical person can do a mental task with less than one second of thought, we can probably automate it using AI either now or in the near future.” If some jobs disappear along the way, others, such as data scientist, will be created. And the movement is well underway. Profiles such as data scientist which barely existed five years ago are superceding repetitive, rote tasks such as data entry.

7 - As a rule of thumb, the higher the cognitive demand, the lower the risk of job displacement

Jobs requiring high levels of social interaction, creativity or strategic cognitive work, or high dexterity in unstructured environments, are at lower risk of short term elimination. See the full article for examples.

8 - Actual job losses will likely be much smaller than the pessimistic forecasts.

Some forecasts put about 9% of jobs in the USA and Europe at risk. PWC researchers predict that 38% of jobs in the US could be at risk of automation by the early 2030s. The actual replacement is likely much lower, around 10-15%.

9 -The AI/Human interplay can deliver 5 business benefits:

AI helps humans perform repetitive tasks, analyze huge data sets and handle routine cases.

  1. Flexibility (Robotics in Auto-manufacturing, Software to improve Product design, Software development estimates)
  2. Speed (Fraud detection, aggregating patient data in cancer treatment, video analytics that enhance public safety)
  3. Scale (Automated applicant screening in recruitment, customer service bots, monitoring systems)
  4. Decision-Making (diagnostic applications in equipment maintenance, real time Robo-advisors in financial services, disease prediction)
  5. Personalization (wearable AI improving the guest experience, wearable sensors in healthcare, AI analytics in fashion retail).

10 - Machines may know something you don’t.

As an executive making decisions in an uncertain environment, you need to focus on the consequences you want (which you know) rather than their probability (which you likely don’t). AI deep learning machines may be able to help you make smarter decisions by providing clues as conditional probabilities, reducing that uncertainty. They can fill information gaps, finding or revealing patterns invisible to you.

See the full article for our model.

11 - Moravec’s paradox explains why no robot can tie its shoes quite like you can.

According to this paradox, high level reasoning requires very little computation, whereas low-level sensorimotor skills require enormous resources. This is why robots are great at precision-welding or rapidly calculating distances in self-driving cars, but still can’t can’t tie their own shoes as fast and fluidly as a human.

12 - AI systems are only as good as the (often poor) data they’re trained on.

In AI kindergarten, the old adage: ‘Garbage in, garbage out’ becomes ‘biases in, biases out’. Not only must algorithms be unbiased, the training data must also be free from a distorted, or unethical, perspective. And of course, the European General Data Protection Regulation (GDPR) requires firms to explain how data are used and kept private. This “right to explanation” gives consumers the right to question firms on the use of their data.

13 - AI involves a trade-off between ‘accuracy’ versus ‘explainability’.

If a deep-learning system provides high predictive accuracy, it may have serious difficulty explaining how its results were derived, turning the algorithm into a black box. In the case of healthcare and consumer-facing applications, these AI will face considerable regulatory scrutiny. So the requirement for more explainability must take into account not just technological issues but financial, legal, ethical and other key considerations.

How will an autonomous car share the road with pedestrians, human-driven vehicles and other autonomous cars? How should conflicts between human values and autonomous navigation systems be resolved? We’ll need to think about robo-ethics, ensuring AI-driven vehicles are well coordinated and aligned with human value systems if they are to enhance, and not harm, our quality of life.

14 - Countries that currently rely on low-cost labor may well lose a competitive advantage.

In China, for example, Foxconn is leading in automating the jobs of blue collar workers. Whether factories in ‘developing’ nations remain a low cost and competitive link in the global supply chain will depend on their success in integrating AI. And government has a big role to play.

15 - Robo-ethics remains a burning platform, which only wise leadership can solve.

How will an autonomous car share the road with pedestrians, human-driven vehicles and other autonomous cars? How should conflicts between human values and autonomous navigation systems be resolved? We’ll need to think about robo-ethics, ensuring AI-driven vehicles are well coordinated and aligned with human value systems if they are to enhance, and not harm, our quality of life.

Conclusion:

Human cognition has a rich host of qualities.

We are flexible. We can adapt quickly to new situations. We can easily adjust the way we interpret information, solve problems, exercise judgment and act to suit our specific context. We possess imagination, intuition, creativity and empathy. Our ability to resolve ambiguous problems and exercise judgment in difficult cases can’t be matched, let alone replaced, by AI. And this doesn’t look set to change any time soon.

Wise decision-making means optimizing the human/machine symbiosis.

Embracing the efficiency and effectiveness of AI/big data analytics, whilst emphasizing human creative intuition, skill and experience.

Just as in any other form of talent management it’s about putting the right mind on the right job, as we welcome AI to the workforce.