Although artificial intelligence is currently making encouraging progress, it has yet to cause revolutionary changes in many industries. In many cases, the problem isn’t necessarily the technology, but the way people perceive it.
"Power and Prediction" is a new book written by an artificial intelligence expert that explores the fundamental challenges of applying artificial intelligence technology in different industries. A sequel to their critically acclaimed Prediction Machines, this book discusses what businesses need to change before they can benefit from the full potential of advances in artificial intelligence.
From point solutions and applications to artificial intelligence systems, industry experts examine the successes and failures of artificial intelligence in different fields. They also provide important insights from past technological revolutions and show how rethinking and designing AI systems from the ground up can help create real value based on powerful machine learning and deep learning algorithms.
Today’s AI systems are predictive machines, meaning they can predict what will happen in the future based on past data. This is what every mathematical model does. But thanks to the availability of large amounts of data and computing, as well as advances in deep learning algorithms, people have been able to create models that can make predictions about complex information such as images, text, and multidimensional data.
In the book "Power and Prediction", the author divides the value of artificial intelligence into three categories: point solutions, application solutions, and system solutions.
So far, most of what people have seen are point solutions and application solutions. These AI systems replace tasks that previously required prediction. For example, in financial services, one of the tasks is to predict which transactions are fraudulent. A machine learning model trained on the right data can take over this task. Point solutions are the low-hanging fruit of artificial intelligence because adopting them requires only minimal investments and changes to the underlying systems.
Another example of a point solution is analyzing radiology scans. There are now several deep learning models that can detect various diseases from X-rays and MRI scans at a level comparable to experienced radiologists.
They are automating one of the many tasks performed by radiologists without requiring any changes to the underlying patient care system.
Artificial intelligence systems can provide greater value by automating new tasks and problems that are not solved by current applications and systems. However, AI systems require a blank slate approach, in which entire processes, workflows, and applications need to be redesigned to solve not only existing problems but also new ones. In order for them to work, AI systems often require new organizational structures and alignment of goals and incentives. This makes AI systems more difficult and risky, but also more rewarding.
The author of "Power and Prediction" writes: "System solutions are often more difficult to achieve than point solutions or application solutions because AI-enhanced decisions affect other decisions in the system. Point Solutions and applied solutions often lead to disruption by reinforcing existing systems, while system solutions disrupt existing solutions. However, in many cases, system solutions may bring the greatest benefits to AI investments The overall return of artificial intelligence."
In the book "Power and Prediction", the author believes that we are now in the "intermediate era" of artificial intelligence. After witnessing this After the power of technology, before its widespread adoption. This is why point solutions are currently a more attractive and popular use case for artificial intelligence.
This has historical precedent. For example, in the late 19th century, when electricity began to be industrialized, its first applications were point solutions. For factories, this means replacing steam engines with electric motors to reduce energy costs. Changing the source of electricity does not require redesigning the factory.
However, the real value proposition of electricity is to decouple the machine from the power source. This enabled new factory designs that were not possible under steam power and made them more productive and less expensive. But that adoption took decades because it required fundamental changes, a break in habits, and an upfront investment that existing businesses were unwilling to make. Those entrepreneurs who seized the opportunity succeeded in taking leading positions and capturing a large portion of the market that later replaced the old market.
One can see these changes in many other industries, such as the rise of online shopping, the advent of personal computers, and the shift from print to digital media.
Artificial Intelligence is an infrastructure technology whose impact technology leaders have compared to electricity. Therefore, this requires a new mentality and bold exploration.
The author of "Power and Prediction" writes: "AI-driven industry transformation takes time, and it is not obvious how to do it at first. Many people may try and fail because they misunderstand the need, or they cannot Let the unit economics work. Eventually, someone will succeed and build a path to profitability. Others will try to imitate. Industry leaders will try to protect their advantage. Sometimes it succeeds. Regardless, the industry will transform, as it always has There will always be winners and losers."
The author of "Power and Prediction" said, "When there is nothing, you will not give up. If there is no need By using the information to make informed choices, you can avoid the consequences of doing things blindly. So it’s not surprising that when AI predictions emerge, the opportunities for their use are not obvious. Potential decision-makers can’t make decisions without this information. There’s a scaffolding built on it.”
Opportunities in artificial intelligence are difficult to spot because they are often hidden behind strict rules and procedures that work well and have been established for a long time. These rules make up for the lack of information. They enable people to make decisions without being able to predict accurate outcomes. They help build systems that, while not optimal, work reliably in many situations.
The key to finding these opportunities is, first, to understand the power of prediction machines, and second, to find where predictions can supersede established rules. A very interesting example that the author explores in the book is the use of artificial intelligence in education.
Thanks to machine learning algorithms and historical data, it is possible to predict student performance, where they will excel and where they will struggle. This gives us the opportunity to provide more personalized content for each student.
But these predictive models are not very helpful in the current education system, which is based on age-based curriculum with only one teacher per class. This system was created because teachers have no way to accurately measure students’ individual learning abilities through their educational trajectories.
To be able to take full advantage of machine learning, people need to rethink the education system in a new way. This new system will replace age-based curriculum with personalized discussions, group projects and teacher support, creating a greater impact on overall education and personal growth and development.
The authors of "Power and Prediction" write: "Age-based curriculum rules are the glue of the modern education system, so artificial intelligence to personalize learning content can only provide limited benefits in this system . To unlock the potential of AI for personalized education, the main challenge is not to build predictive models, but to decouple education from the age-based curriculum rules that currently glue the system together."
Successful applications of artificial intelligence require what the author of "Power and Prediction" calls "systems thinking," which contrasts with "task thinking." A task mindset focuses on cost savings. Systems thinking focuses on value creation. The task mindset focuses on automating individual tasks. Systems thinking recognizes the need to rebuild systems that generate value based on machine predictions and human decision-making.
People have already seen this happening in some industries and large technology companies such as Amazon and Google, which have formed profit-making systems based on artificial intelligence predictions to recommend personalized content.
Perhaps one of the important elements of the systems mindset is the power shift that occurs with the adoption of artificial intelligence. As the system changes, so do the people with decision-making power.
The author of "Power and Prediction" writes, "While artificial intelligence cannot turn decisions over to machines, it can change who makes decisions. Machines have no power, but once deployed, they can change who has People with power. When machines change decision-makers, the underlying systems must change too. The engineers who build the machines need to understand the consequences of the judgments they embed into the products. Those who used to make decisions in the moment may no longer need to."
A hypothetical example that the authors explore in the book is heart attack risk. Currently, this risk assessment is done through testing in hospital, with the decision made by the specialist who carries out the test.
Hypothetically, it is possible to build an artificial intelligence system that predicts heart attack risk based on data collected by wearable devices such as smart watches. It would then be possible to move these predictions from the triage space of a hospital emergency department to the patient's home. In this case, many patients will never need to go to the hospital after being diagnosed with a condition that a pharmacist or primary care physician can help treat at home.
No matter where one stands on the scientific and philosophical debates surrounding artificial intelligence, what one can agree on is that predictive machines have a lot to offer and are only scratching the surface. Leveraging their full potential starts by going back to the drawing board and rethinking how people would design systems if they had the ability to predict. ?
The above is the detailed content of Why applying artificial intelligence requires a major mindset shift. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
