The goal of intelligent decision-making is to turn information into better action.
Editor’s note: Data is beautiful, but the most important thing is decision making. Decision making is the choice of any entity between options. Studying AI requires an in-depth understanding of decision-making intelligence. We can often consider quantitative and qualitative perspectives. Quantitative mainly involves data science, which is qualitatively related to society and management science. This article was compiled from Medium, by Cassie Kozyrkov, the original title “Introduction to Decision Intelligence”.
On the prairie, some people think about avoiding lions; when they return to the real world, some people want to be AI leaders and want to design a data warehouse: What are the similarities between the two? They all involve “Decision intelligence.”
Decision intelligence is a new academic theory that involves all aspects of choice. Decision-making intelligence science integrates data science, social science, and management science to help people use their data to improve their lives, optimize their businesses, and change the world. As the AI era approaches, decision intelligence becomes more and more important.
In short, the goal of intelligent decision-making is to turn information into better action, regardless of size.
Today we briefly introduce the basic terms and concepts of decision intelligence science.
What is a decision
Data is beautiful, but the most important thing is decision making. Data runs through our decisions, through our actions, and affects the world around us.
What is “decision”? We believe that decision making is the choice of any entity between options.
The computer system marks the picture, pointing out whether the animal in the picture is a cat. This is a decision; the human leader in charge of the project makes a decision to see if the system is launched, which is also a decision.
What are the decision makers
The “decision makers” we are talking about here are not the stakeholders or investors who control the project team, but the people who influence the decision-making structure and the situational framework. In other words, he is well-designed and is a creator rather than a destroyer.
Decision making
The term “decision making” is used in many disciplines and is used in a different sense. It may mean something like this:
– Take action with other alternatives (from this level, the decision may be made by a computer or lizard).
—Fulfill the functions of human decision makers, and human decision makers take responsibility for decisions. Although computer systems can make decisions, we can’t call them “decision makers” because they don’t take any responsibility for the results, and ultimately the responsibility falls on the shoulders of humans.
Decision Intelligence Classification
To understand the intelligentization of decision-making, it can often be considered from the perspective of quantitative and qualitative. Quantitative mainly involves data science, and qualitatively involves society and management science.
Qualitative: Decision Science
The science involved in characterization is what we call decision science, and decision science needs to answer the following questions:
——How to set decision criteria and develop indicators. (all)
– Is the indicator you choose compatible with the incentive? (economic)
——What quality should be used to make decisions? What price should I pay for perfect information? (Decision Analysis)
—What role do emotions, inspirations, and prejudices play in decision making? (psychological)
——How do biological factors such as cortisol levels affect decision making? (Neuroeconomics)
——What impact does the change in information representation have on selective behavior? (Behavioral Economics)
——When making decisions in a group environment, if you optimize the results? (Experimental Game Theory)
After designing a decision-making environment, how to balance between massive limits and multi-level goals? (design)
-Who will feel the consequences of this decision? What kind of experience will different groups get? (UX Research)
——Is this decision goal ethical? (philosophy)
This is just a little bit of a little bit, there are a lot!
Brain
Human beings are not optimizers, but satisfied people, that is to say, human beings are satisfied with “good enough” rather than “perfect”.
Returning to reality, humans use cognitive heuristics to save time and effort. Many times, this is good. If you encounter a monkey on the prairie, we can run for the first time, not carefully calculated and then act. “Satisfaction” can also save energy. The human brain is like an electronic device with very high energy consumption. Although it weighs less than 3 pounds, it uses nearly one-fifth of the energy.
99.9% of people don’t face the threat of lions every day, because “stolen work”, and ultimately our brains are not fully prepared for the modern environment.
Understand the decision-making mechanisms of the human brain and understand how the human brain translates information into action so that we can protect ourselves from brain defects. We can also develop tools based on understanding, enhance human capabilities, and make the environment more harmonious with the brain.
If you think that AI will eliminate humans, then you have to think about it for a while. All technologies are a reflection of the creators. Large-scale systems can amplify human defects. For this reason, if you want to be a responsible AI leader, you must improve your decision-making intelligence.
Maybe you didn’t make a decision
Sometimes, when you think about your own decision criteria, you will find that nothing in the world changes your mind. You have chosen your own actions. Now you just want to make yourself feel more comfortable. This kind of understanding is beneficial because it prevents you from wasting time and helping you to re-adjust bad emotions. When you do anything, the data really adds a lot of annoyance.
Unless you encounter an unknown fact and need to take different actions, there is no need to make a decision. Of course, accepting some decision analysis training does allow you to see things more clearly.
Make a decision under perfect information
Nothing is better than the facts. When we make decisions, if there are facts that can be relied upon, decisions are often made based on facts. Because of this, our first priority is to figure out how to deal with the facts.
What can be done with facts
– You can use facts to make very important decisions. If the decision is really important, you will rely on qualitative factors to help you make decisions.Make decisions more sensible. Psychologists know that information can manipulate you in ways that you don’t want to accept, so they offer a lot of advice to help you pick up the information—pre-received information before making a decision.
——You can use your facts to consolidate your point of view. (From “I am looking forward to the sun outside!” becomes “I know the outside is sunny”)
——You can use facts to make an important decision based on existence. “I found an Ebola case next door, and I have to leave quickly.” This is a decision based on existence. The reasoning based on the existence of the decision is because of the facts that were previously unknown. It has shaken your methods and methods. After you discover that the environment in which you make decisions is not solid, it is a hasty construction.
– You can use facts to “automate” a lot of decisions.
– You can use the facts to create an automated solution. You can observe the facts about the system and then write the code based on the observations. Traditional methods are programmed based on thinking rather than information, and the above methods are better.
——You can use the facts to get the best solution, solve the problem perfectly and solve it automatically.
—You can use the facts to inspire yourself and tell yourself how to make important decisions in the future.
– You can use facts to evaluate what is being done.
—You can use facts to make certain decisions that are not so important, that are less serious, and that you don’t need to be too serious when making decisions, such as: “What do you eat for lunch today?” If all decisions are pursued Perfection will eventually lead to failure to reach the standard in life, and you will be kidnapped by meaningless perfectionism. But don’t rely too much on this approach. If you want to improve the quality of your decisions, you still need more stringent requirements.
Training yourself with decision science, when we need to make decisions based on existence, we can save energy. In other words, we can improve the overall decision quality with the same energy. This method is quite practical, but it is really difficult to do.
Data collection and data engineering
If we can grasp the facts, we have already mastered them. In today’s world, we are busy with information. Data engineering is a complex science. If it is deployed on a large scale, information must be reliable. This is the mission of data engineering. It is easy to go to the store to buy an ice cream. If important information is listed in the form of a spreadsheet, data engineering is equally easy.
But if you let 2 million tons of ice cream, you still have toIt is not easy to guarantee that it will not melt. If you want to design, build, and maintain a huge warehouse, and you don’t know what to store next, it may be fish, or it may be awkward, which is even more difficult.
For decision intelligence, data engineering is only a sibling and a key collaborative discipline; data science contains a lot of traditional expertise, and we use it to design and manage facts.
Quantification: Data Science
When you make a decision, look at all the facts you need, use a search engine or analytics to get the facts, and all that’s left is execution. You do not need data science when you are done.
If the facts given are not ideal facts that can help you make decisions? If it is only part of the facts? Maybe what you need is the fact of tomorrow, but what you have is the facts of the past, what should I do?
Maybe you want to know what all potential users think about your product, but you only ask a few hundred people. As a result, there is uncertainty when you make decisions. What you know is not what you want to know. How to do? Let data science help you.
Obviously, when you find that the reality you have is not the fact you need, you need to modify the method strategy. Maybe they are just small puzzles in big puzzles, maybe they are wrong puzzles, but you can’t find better ones.
Data science becomes interesting when you can’t help but beyond data, but be careful not to make the same mistakes as Icarus (Icarus’s father’s work The wings fled from Crete, and the wax used to stick the wings melted because it was too close to the sun.)
In short, we can draw wisdom from the previously closed disciplines, and use these wisdom to improve decision-making efficiency. The essence of decision-making intelligence is here. We have a variety of perspectives on decision-making intelligence, and intelligent decision-making combines these ideas to make us stronger and more solid, ultimately helping humans break the traditional limits.
We can use the kitchen AI to compare. If the purpose of AI research is to develop a new microwave oven and put the AI into the microwave oven, then the intelligent decision-making mission is to make the microwave oven safe to achieve the goal, and can be used when no microwave oven is needed. Other things. Goals are often the starting point for intelligent decision making.
Translator: Xiaobing Hand