Original Source: AI New Intelligence

Author: Bill Gates

Image Source: Generated by Wujie AI

Since co-founding Microsoft with Paul Allen, my passion for software remains undiminished.

However, despite significant progress in software over the past few decades, in many ways, software still seems quite "clumsy."

For any task on a computer, you need to specify the application to use. While you can use Microsoft Word or Google Docs to write a business proposal, these tools cannot help you send emails, share selfies, analyze data, schedule gatherings, or buy movie tickets. Even the best websites can only partially understand your work, personal life, interests, and relationships, and have limited ability to use this information to serve you. Currently, this work is usually done by humans, such as your close friends or personal assistants.

But in the next five years, all of this will change completely. You will no longer need to switch between different applications for different tasks. You just need to tell your device in ordinary language what you want to do. Software will be able to respond in a personalized way based on the amount of information you are willing to share, as it has a deep understanding of your life.

In the near future, anyone on the internet will be able to have a personal assistant powered by advanced artificial intelligence.

These types of software, capable of understanding natural language and completing various tasks based on their understanding of the user, are called "Agents." I have been thinking about Agents for nearly 30 years. I mentioned them in my book "The Road Ahead" published in 1995, but only recently, due to the progress of artificial intelligence, have they become truly practical.

Agents will not only change the way people interact with computers, but also revolutionize the software industry, triggering the biggest revolution in the field of computing since we moved from typing commands to clicking icons.

1 Everyone's Personal Assistant

Critics have pointed out that software companies have previously offered similar services, but users did not fully embrace them. (People still make jokes about Clippy, the digital assistant we once added to Microsoft Office and later abandoned.) So why would people use Agents?

The answer is that Agents will be greatly improved. You will be able to have more nuanced conversations with them. They will become more personalized and not limited to relatively simple tasks such as writing letters. The difference between Clippy and Agents is as great as the difference between rotary phones and smartphones.

Agents will be able to assist you in all activities if you are willing. Once given permission, Agents can track your online interactions and actual location, gaining a deep understanding of the people, places, and activities you are involved in. They will understand your personal and work relationships, interests, preferences, and schedule. You can choose when and where to let them intervene to help you make decisions or complete tasks.

To understand the huge changes Agents will bring, we can compare them to today's AI tools. These tools are mostly robots, limited to individual applications, and usually only intervene when you make specific requests. Because they cannot remember your usage habits, they cannot learn and adapt to your preferences.

"Clippy is just a robot, not an Agent."

Agents, on the other hand, are more intelligent. They can make proactive suggestions, not just react when you make a request. They can cross applications to complete tasks, and over time, they will continue to improve as they can remember your activities and identify intentions and patterns in your behavior. Based on this information, they will proactively offer what they think you need, although the ultimate decision-making power always remains in your hands.

Imagine planning a trip. A travel robot may only be able to help you find a hotel within your budget. But an Agent will know when you are traveling and, based on its understanding of whether you always try new destinations or like to revisit the same place, can suggest destinations for you. When asked, it will also recommend activities based on your interests and adventurous tendencies, and book your favorite restaurants for you. If you want this kind of deeply personalized planning today, you need to pay for a travel Agent and spend time telling them your needs.

The most exciting impact of AI Agents is that they will make services that are currently too expensive for most people become more accessible. They will have a huge impact in four areas: healthcare, education, productivity, and entertainment and shopping.

2 Healthcare

In today's healthcare field, artificial intelligence mainly assists in handling administrative tasks. For example, systems like Abridge, Nuance DAX, and Nabla Copilot can record audio during a doctor's diagnosis and organize it into notes for the doctor to review.

The real change will come when Agents can assist patients in preliminary triage, provide advice on handling health issues, and judge when it is time to seek medical care. These Agents will also help healthcare workers make wiser decisions and improve their efficiency. (For example, applications like Glass Health can analyze patient summaries and suggest possible diagnoses to doctors.) This assistance to patients and healthcare workers is particularly important for people in poor countries, where many cannot see a doctor at all.

The promotion of these clinical Agents will be slower than other types, as accuracy is a matter of life and death. People need to be sure that while these health Agents are not perfect and may make mistakes, they are generally beneficial. Of course, humans make mistakes too, and lack of healthcare is also a problem.

"In the United States, half of the veterans who need mental health care do not receive the treatment they need."

Mental health care is another area where Agents will be a service that is almost universal. Today, weekly therapy seems like a luxury. But in reality, there is a lot of unmet need, and many people who need therapy cannot get it. For example, a study by the RAND Corporation found that in the United States, half of the veterans who need mental health care do not receive the treatment they need.

AI Agents trained in mental health will make therapy more affordable and accessible. Wysa and Youper are two early chatbot examples in this field. But the role of Agents will be more profound. If you are willing to share sufficient information with a mental health Agent, it will be able to understand your life history and relationships. It will be on call whenever you need it, never getting impatient. With your permission, it can even monitor your physiological responses during therapy through a smartwatch—for example, when your heart rate increases when discussing issues with your boss—and advise you when to seek help from a human therapist.

3 Education

For decades, I have been hopeful about how software can simplify the work of teachers and help students learn. It will not replace teachers, but will complement their work—tailoring teaching materials to students and freeing teachers from tedious paperwork and other tasks, allowing them to focus more on the most important parts of their work. These changes have finally begun to materialize in a significant way.

The current advanced technology is the text-based robot Khanmigo developed by Khan Academy. It can provide tutoring for students in mathematics, science, and humanities—for example, explaining quadratic equations and creating math problems for practice. It can also help teachers with curriculum planning and other tasks. I have long been a supporter of Sal Khan's work and recently invited him to discuss education and AI on my podcast.

But text-based robots are just the first step—Agents will open up more learning opportunities.

For example, few families can afford one-on-one tutoring to supplement classroom learning for students. If Agents can mimic the characteristics of effective tutoring, they will provide this supplementary teaching for everyone who needs it. If a tutoring Agent knows a child likes Minecraft and Taylor Swift, it will use Minecraft to teach volume and area calculations and teach storytelling and rhyme skills using Taylor's lyrics. This experience will be richer than today's text-based tutoring—combining graphics and sound—and more personalized.

4 Productivity

In this field, the competition is already very fierce. Microsoft is integrating its Copilot feature into services like Word, Excel, and Outlook. Google is doing similar things through Assistant, Bard, and its productivity tools. These intelligent assistants can perform many tasks, such as converting text documents into slides, answering spreadsheet-related questions in natural language, and summarizing email discussions while reflecting everyone's viewpoints. The capabilities of Agents will be even more powerful. Having an Agent is like having someone dedicated to assisting you with various tasks and independently completing them when you need. For example, if you have a business idea, the Agent can help you write a business plan, create presentations, and even generate images of what your product might look like. Companies can provide Agent services to employees, allowing the Agent to directly participate in every meeting and answer questions. "Your Agent will proactively suggest sending flowers if your friend just had surgery, and can handle the flower ordering for you." Whether you are in the office or not, your Agent can provide assistance just like today's personal assistants. If your friend recently had surgery, your Agent will proactively suggest sending flowers and can help you place the order. If you want to reunite with college roommates, it will collaborate with the other person's Agent to arrange a meeting time and remind you before you arrive that their eldest child has just started college locally. **5 Entertainment and Shopping** Today, artificial intelligence can help you choose a new TV, recommend movies, books, shows, and podcasts. Similarly, a company I have invested in recently launched Pix, which allows you to ask questions (e.g., "What Robert Redford movies would I like? Where can I watch them?"), and then makes recommendations based on your past preferences. Spotify's AI-driven DJ not only plays songs based on your preferences but also interacts with you, and can even address you by name. Agents will not only make recommendations but also help you act on those recommendations. If you want to buy a camera, you can have the Agent read all the reviews, summarize them, provide recommendations, and place the order for you once you make a decision. If you tell the Agent you want to watch "Star Wars," it will know if you are subscribed to the relevant streaming service, and if not, it will suggest signing up. If you are unsure about what to watch, it will provide customized recommendations and then help you play the selected movie or show. You can also get news and entertainment content tailored to your interests. CurioAI is an example of this trend, as it can create custom podcasts based on any topic you request. **6 The Massive Changes in the Tech Industry** In short, Agents will be able to help with almost all activities and aspects of life. This will have a profound impact on the software industry and society as a whole. In the computing industry, we talk about platforms—the foundational technology on which applications and services are built. Android, iOS, and Windows are examples of platforms. Agents will become the next platform. "To create a new application or service, you just need to tell your Agent your requirements." To create a new application or service, you do not need to have programming or graphic design skills. You just need to tell your Agent your requirements. It will be able to write code, design the look and feel of the app, create a logo, and publish the app to online stores. The GPTs released by OpenAI this week provide a glimpse of the future, where non-developers can easily create and share their own assistants. Agents will change the way we use software and how software is developed. They will replace search engines because they will be more efficient in finding and summarizing information for you. They will replace many e-commerce websites because they can find the best prices for you and are not limited to a few vendors. They will replace word processing software, spreadsheets, and other productivity applications. Today's separate businesses—search advertising, ad-supported social networks, e-commerce, productivity software—will merge into one business. I do not think any single company will dominate the Agent business—there will be many different AI engines to choose from. Today, Agents are embedded in other software, such as word processing and spreadsheets, but eventually, they will operate independently. While some Agents will be free to use (and supported by ads), I believe you will need to pay for most Agents, which means companies will have an incentive to have Agents represent your interests rather than advertisers. The number of companies researching AI this year suggests that there will be exceptionally fierce competition, making Agent prices very affordable. But before the advanced Agents I described become a reality, we need to address many questions about technology and its use. I have previously written about the issues raised by AI, so here I will focus specifically on Agents. **7 Technological Challenges** The data structure of Agents has not yet been figured out. To create a personal Agent, we need a new type of database that can accurately capture subtle differences in your interests and relationships and quickly extract this information while protecting privacy. We have begun to see some new ways of storing information, such as vector databases, which may be better suited for storing data generated by machine learning models. Another mystery is how many Agents people will interact with. Will your personal Agent be separate from your therapy Agent and math tutoring Agent? If so, in what situations do you want them to work together, and when should they remain separate? "How you interact with your Agent is currently being explored by companies, including options such as apps, glasses, pendants, pins, and even holograms." How will you interact with your Agent? How you interact with your Agent is currently being explored by companies, including options such as apps, glasses, pendants, pins, and even holograms. These are all possible ways, but I believe the first major breakthrough in human-Agent interaction will be through earphones. If your Agent needs to contact you, it will speak to you through earphones or display information on your phone. ("Your flight is delayed. Do you want to wait, or do you need me to help you rebook?") If you are willing, it can also monitor the sound entering your ear and enhance the auditory experience by eliminating background noise, enhancing difficult-to-understand speech, or making it easier to understand people with heavy accents. There are other challenges. There is currently no standard protocol allowing Agents to communicate with each other. The cost of Agents needs to be reduced to make them affordable for everyone. The process of activating an Agent to get the right answer needs to be made simpler. We need to guard against illusions in fields like healthcare, where accuracy is crucial, while ensuring that Agents do not harm people due to bias. We do not want Agents to do things they should not do. (Although my concern about malicious Agents is less than my concern about human criminals using Agents for malicious activities.) **8 Privacy and Other Major Issues** As these technologies converge, online privacy and security issues will become even more urgent than they are now. You will want to be able to decide what information your Agent can access, to ensure that your data is only shared with the people and companies you choose. But who owns the data you share with the Agent, and how can it be ensured that it is used appropriately? No one wants to start receiving related ads because they told their therapy Agent something. Can law enforcement use your Agent as evidence against you? When will your Agent refuse to do something that may be harmful to you or others? Who determines the embedded values of the Agent? Another question is how much information Agents should share. Suppose you want to meet a friend: if your Agent communicates with their Agent, you do not want it to say, "Oh, she's meeting other friends on Tuesday and doesn't want you to join." If your Agent helps you write work emails, it needs to know not to use your personal information or proprietary data from previous jobs. Many of these issues are already the focus of the tech industry and legislators. I recently attended an AI forum organized by Senator Chuck Schumer, where other tech leaders and many U.S. senators were present. We shared our thoughts on these and other issues and discussed the need for strong legislation by lawmakers. But other issues will not be decided by companies and governments. For example, Agents may change the way we interact with friends and family. Today, you can show you care by remembering details of their lives, such as their birthdays. But when they know it is likely your Agent reminding you of this and responsible for sending flowers, will it still mean something to them? "In the distant future, Agents may even force humans to face profound questions about purpose." In the distant future, Agents may even force humans to face profound questions about purpose. Imagine a future where Agents become so excellent that everyone can enjoy a high-quality life with almost no work. In such a future, how will people use their time? Will anyone still want to be educated when Agents have all the answers? Can a safe and prosperous society be maintained when most people have a lot of free time? But we are far from that point. In the meantime, Agents are quietly arriving. In the coming years, they will completely change our way of life, both online and offline.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。