Home International How to Move up When the Only Way Is Down: Learning from...

How to Move up When the Only Way Is Down: Learning from AI

0

This is an excerpt from How To Move Up When The Only Way is Down: Lessons from Artificial Intelligence for Overcoming Your Local Maximum, in which Judah Taub shares insights into how humans can achieve better decision-making to surpass expectations by learning from the way AI overcomes local maximums.


Consider the following real-life scenarios:

  1. The manager of an English football team at the bottom of the second division. 

All the team players are average except for the star striker, who is responsible for most of the team’s goals. The fact that all the other players are centered around the star player seriously limits their play and their own development. In the long run, the team would be better off without the star player. In the short term, there is a price to be paid: the team will likely go down a division, and it could take years to recover.

  1. The military needs to determine how to spend their budget. 

Combat divisions need ammunition and motor vehicles, and they need to invest in intelligence to predict the type of warfare anticipated. How do you trade-off building the military force (running up the mountain) while also balancing intelligence to make sure you are investing in the appropriate tools and training (heading in the right direction)?

  1. The CEO of a successful start-up that has gained tremendous traction.

Out of the gate and on a shoestring budget, the CEO introduced an immediately popular and widely adopted freemium product, generally known to be the envy of his heavily backed competitors. However, she needs to raise more money to bring the product to a broader market. The investors are advising her to prioritize short-term revenues, which means sacrificing part of her unique brand and potentially alienating her original community of supporters.

  1. A senior government official charged with upgrading national infrastructure.

New 5G telecom technology promises major benefits throughout the country’s economy. While it is clear 6G and 7G technologies will arise in the future and may render the enormously expensive investments in 5G redundant before too long, voters are hungry for speedy results. How do you balance the huge potential without getting stuck with a huge “sunk cost”?

Local Maximum offers a simple framework to understand why some businesses plateau, why some people find themselves in jobs they can’t leave, and why we find ourselves trapped in situations that prevent us reaching our full potential in so many fields of life. Understanding this concept gives us the tools to ask:

  • What are the behaviors or decisions that lead us to a Local Maximum?
  • What can we do to steer ourselves away from these limiting Maximums before we get there?
  • And, if we do get there, what can we do to get unstuck?


A Prime Example: The Delivery Route

A classic example of the Local Maximum challenge is Amazon Prime and its complex system to manage deliveries. Consider how the system determines the most efficient route for the driver to deliver packages to hundreds of locations around a city. This may sound like a simple A to B mapping project, but finding the optimal solution is nearly impossible due to the sheer volume of options.

Think about it this way. Imagine you need to make 10 deliveries across the city in a day. How many possible optimal routes are there? (The answer is over 3M!) Now, pretend you have to make 20 deliveries, that is 3+10^64 optional routes. (That’s more than the number of steps it would take to “walk” to the sun!) In reality, Amazon has thousands of drivers, and each of them make hundreds of deliveries a day; the number of route options is simply too large for the mind to comprehend. More so—and this might come as a surprise—the number of route options is too large for even the fastest and best computer to comprehend. So, how do computer scientists overcome this? They turn the problem into mountains.

So, consider Amazon Prime as a mountain climber:

Amazon Prime delivers packages. Its profit relates directly to the speed of its deliveries. The more deliveries it can make in an hour, the more profit. The process of planning delivery routes is a mountain that must be climbed. To solve the task, the data scientist converts the deliveries into a topographic map: the better the delivery route, the higher the point it represents on the map. (Routes that are similar appear next to each other.) Next, the data scientist asks himself: how do I reach the route/peak of greatest efficiency and avoid the costs of adopting a route/peak that looks efficient, but that ignores faster, more cost-effective routes/peaks?

The Amazon Prime solution, represented by the figure, as if on a desert field. Each point on the field is a different potential solution, with the height representing the number of deliveries per hour the driver can make at that point. Notice how there are points where the algorithm can’t improve with only one simple step, such as the 25 deliveries per hour point the current Amazon algorithm is heading towards. Hence, they are Local Maximums the system may return as the suggested solution.

We earn a commission if you make a purchase, at no additional cost to you.

Amazon Prime, and many other businesses, have spent huge sums of money and dedicated their brightest minds to develop solutions and new logics to alleviate the challenge of a Local Maximum. Until recently, humans have not had the tools to address such dilemmas, or to even think about them effectively. But now that billions of dollars have been poured into improving computers’ abilities to limit those effects, it is time for us humans to leverage these learnings so that we, too, can both identify Local Maximums and limit their negative impacts in our personal and professional lives.

Most decisions include an element of Local Maximum, and the more complex the decision, the stronger the effects and dangers of a Local Maximum. This concept can apply to decisions that have small effects, such as which ice cream flavor to choose or which shoes to buy, and to decisions that have very large effects, such as which job to pursue, how to help people out of extreme poverty, how to build a company’s business roadmap, or even how to reach a carbon neutral society. The concept of Local Maximum offers new ways of thinking about human challenges as well as ways to avoid or address those problems, whether it’s global warming or what to order for breakfast.

My work with start-ups and various other life experiences with Local Maximums has helped me to understand we are all in the desert on our personal or corporate journeys, like our paratrooper in training at the top of this chapter, trying to navigate our way to the highest mountaintop. Many times, we know we are not climbing the right mountain, but we are concerned about the costs of going back down. Other times, we may not be aware there is a much better mountain right around the corner. We need to understand our terrain to navigate it most effectively.

This excerpt from How To Move Up When The Only Way is Down: Lessons from Artificial Intelligence for Overcoming Your Local Maximum by Judah Taub, copyright October 2024, is reprinted with permission from Wiley, the publisher.

NO COMMENTS

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Exit mobile version