Bill Gates personally revealed the importance of “OpenAI Spring Festival Gala”

Machine capability report

Editor: Wu Xin

This article was published just after OpenAI’s first developer conference (also called the Spring Festival Gala for developers by some people). Gates obviously also understood the content of the conference. In this article, he gave the most understandable interpretation of where GPT technology will go and how it will affect our lives.

I still love software as much today as I did when I founded Microsoft. The software improved greatly over the next few decades, but was still quite clunky in many ways.

To perform any task on your computer, you must tell the device which application to use. You can use Microsoft Word and Google Docs to draft a business plan, but they can't send emails, share selfies, analyze data, arrange parties, or buy movie tickets for you. Even the best websites don't fully understand your work, personal life, interests, and relationships, so they're limited in what they can do. You can only let a close friend or personal assistant do these things for you

In the next five years, this situation will completely change. Simply tell the device what you want to do, no more calling different apps for different tasks. The software will respond autonomously based on your choices and the information provided by your device as it learns more about your life. In the near future, any netizen will be able to have a personal assistant powered by artificial intelligence, whose functions will far exceed today's technical level

In China, this type of software is called an agent (Agent), which responds to natural language and completes a variety of tasks based on its knowledge of the user. I have been thinking about agents for the past 30 years and wrote about them in my 1995 book The Road Ahead. However, it is only recently, thanks to advances in artificial intelligence, that agents have become practical.

Agents will not only change the way everyone interacts with computers, they will also disrupt the software industry and set off a revolution from The biggest computing revolution since typing commands to clicking icons.

# When Bill Gates' first book came out, people were full of curiosity about the future development of digital technology and its impact on our lives. In the book, Bill Gates shares his perspective and makes some predictions about upcoming breakthroughs in computing technology over the next few decades, specifically the Internet, including intelligent agents. Later, however, Bill Gates admitted that his predictions at the time may have been a little too optimistic. Some critics pointed out that there had been previous predictions. Similar things, but users didn't fully embrace them (like Microsoft Office's digital assistant Paperclip, which was eventually abandoned and became the butt of jokes). So why do people use agents?

"Paperclip", exposing the memory of age

The answer is that an agent would be much better. You can have nuanced conversations with them, and they become more personalized and not limited to relatively simple tasks like writing letters. "Paperclips" have a lot in common with smart bodies, just like landline phones with turntables and smart phones in the past also had many similarities.

As long as you are willing, the intelligent body will help you complete everything. As long as it's empowered to track online interactions and real-world locations, it can fully understand event participants, locations, and the event itself. It can map out your personal and work relationships, hobbies, preferences and schedules. You can decide when and how the agent gets involved, or you can make the decision yourself.

To understand the huge changes that agents will bring, let’s compare some of the current AI tools. Most of them are bots, usually limited to an application and only passively intervening when you enter something specific or ask for help. They don't remember how you behave, so they don't get better or learn. to any preference of yours. Microsoft’s “Paperclip” is a robot (bot), not an intelligent agent.

The agent is smarter. They're proactive, make suggestions before you make them, can complete tasks across apps, and get better over time because they remember your activity and recognize your behavioral intentions and patterns. They will provide you with what you need based on these, although you are the final arbiter.

If you plan a trip, the travel bot will find hotels that fit your budget, but the agent knows more, such as what time of year you usually travel, whether you like to try new places, or whether you prefer to travel back and forth. in the same place and suggest destinations accordingly. It will also recommend things to do and book your favorite restaurants based on your interests and risk appetite. Today, if you want this kind of deeply customized planning, you can only spend money to find a travel agency to achieve it.

The most exciting impact of AI agents is how to significantly reduce the cost of today’s too-expensive social services. It will have a particularly large impact on four areas: health care, re-written content is: education is A very important field that has a profound impact on the development of individuals and society. The purpose of education is to develop students' knowledge, skills and values ​​to enable them to succeed in future life and work. The education system should provide fair opportunities and ensure that every student receives a high-quality education. In addition, education should also focus on cultivating students' creativity, critical thinking and problem-solving skills to cope with increasingly complex social challenges. To achieve these goals, educational policies and teaching methods need to be constantly innovated and improved. Education is not only the responsibility of schools, families and society should also actively participate in and support the development of education. By working together, we can build a more prosperous and progressive society, productivity and entertainment shopping.

Healthcare

Now, the main role of AI in healthcare is to help with administrative tasks. For example, Abridge, Nuance DAX, and Nabla Copilot record medical procedures and write notes for the doctor to review.

The real transformation will come when intelligent agents can perform basic triage, make recommendations and decide whether a patient needs to seek treatment. These intelligent agents will also help medical staff make decisions and improve work efficiency. (Apps like Glass Health can already analyze summaries and provide diagnostic recommendations for doctors to consider.) This ability to assist patients and health care workers is especially important in underdeveloped countries. In these countries, many people have no access to medical services.

However, the rollout of clinician intelligence will be relatively slow, after all, it is a matter of life and death. People need to see evidence that, while imperfect and making mistakes, intelligent agents are generally beneficial to human welfare. Of course humans make mistakes, and lack of access to medical care is also a problem.

Another example is that with an intelligent agent, everyone has a psychiatrist. Today, weekly psychotherapy is a luxury, and many people who could benefit from therapy are not receiving it. For example, the RAND Corporation found that half of U.S. veterans who need psychological treatment do not receive corresponding services.

In this regard, well-trained AI agents will significantly reduce service costs, allowing more people in need to enjoy them. Wysa and Youper are examples of early chatbots. But the agent’s services will go much deeper. If you share enough information with them, they can learn about your life history and relationships, and are always at your disposal without ever getting impatient. If allowed, it could even use a smartwatch to monitor your response to treatment options—such as whether your heart beats faster when you talk to your boss about a problem—and recommend when it's time to see a human psychiatrist.

The rewritten content is: Education is a very important field that has a profound impact on the development of individuals and society. The purpose of education is to develop students' knowledge, skills and values ​​to enable them to succeed in future life and work. The education system should provide fair opportunities and ensure that every student receives a high-quality education. In addition, education should also focus on cultivating students' creativity, critical thinking and problem-solving skills to cope with increasingly complex social challenges. To achieve these goals, educational policies and teaching methods need to be constantly innovated and improved. Education is not only the responsibility of schools, families and society should also actively participate in and support the development of education. By working together, we can build a more prosperous and progressive society

For decades, I have been interested in how software can make the teaching and learning process easier for teachers and students. Software will not replace teachers, it will assist teachers in their work, such as providing personalized guidance to students, helping teachers get rid of paperwork and focus on the most important work content. Now, change is finally unfolding in dramatic fashion.

Currently, Khanmigo, the text-based bot created by Khan Academy, is the most advanced and can tutor learning in math, science and humanities. For example, it can explain quadratic equations and provide related exercises. It can also help teachers with things like writing lesson plans. I have always been a fan and supporter of Khan Academy founder Sal Khan, and he recently talked about AI on my podcast. Rewritten: Education is a very important field for the development of individuals and society. Have far-reaching consequences. The purpose of education is to develop students' knowledge, skills and values ​​to enable them to succeed in future life and work. The education system should provide fair opportunities and ensure that every student receives a high-quality education. In addition, education should also focus on cultivating students' creativity, critical thinking and problem-solving skills to cope with increasingly complex social challenges. To achieve these goals, educational policies and teaching methods need to be constantly innovated and improved. Education is not only the responsibility of schools, families and society should also actively participate in and support the development of education. By working together, we can build a more prosperous and progressive society.

However, text-based robots are only the first wave, and intelligent agents will open up more possibilities. For example, few families can afford "one-on-one" tutoring. If an agent knows how to make tutoring efficient, it can provide tailored tutoring for each student. For example, if the agent knew that a child liked Minecraft and Taylor Swift, it would use Minecraft to teach them volume and area calculations, and Taylor's lyrics to teach them stories and how to rhyme. The experience will be richer (with both images and sounds) and more personalized than today’s text-based bot solutions.

The productivity of individuals and organizations refers to the ability and efficiency of individuals or groups in the production and work process. Increased productivity is essential for individual and organizational development. Individual productivity can be improved by upgrading skills, enhancing expertise and improving work methods. The productivity of an organization can be improved by optimizing processes, rationally allocating resources and cultivating team spirit. Improving the productivity of individuals and organizations not only improves work efficiency and quality, but also brings more opportunities and competitive advantages to individuals and organizations. Therefore, both individuals and organizations should pay attention to and continuously strive to improve productivity

Competition in this field is fierce. Microsoft is integrating Copilot into Word, Excel, Outlook and other services. Google is doing something similar. These Copilots can do things like convert written documents into slides, use natural language to answer questions about spreadsheets, and summarize email threads from different people's perspectives.

Agents will do more. It's like having someone specialize in helping you complete various tasks. You have a business idea, and the agent will help you write a business plan, create a presentation for it, and even generate images showing what the product might look like. Companies will have agents that attend every meeting and provide counsel to employees.

Whether working in the office or not, agents will be able to provide support to business executives in the same way that today’s personal assistants do. If your friend has just had surgery, the smart phone will proactively suggest sending flowers and help you order . If you want to catch up with your college roommate, it will discuss the party time with the college roommate's agent and remind you when you go to the appointment that their eldest child has just been admitted to the local university.
Entertainment and Shopping

AI can already help you pick out new TVs and recommend movies, books, shows and podcasts. Likewise, a company I invest in recently launched Pix, which can answer your questions (eg, I like Robert Redford movies, where can I see them?) and give you suggestions based on your preferences. Spotify also has an AI-powered DJ that can not only play songs based on your preferences, but also communicate with you and even call you by your name.

The agent will not only simply make suggestions, but also help you take action. If you want to buy a camera, you can let the robot browse the reviews for you, summarize and make suggestions, and help you place an order after you decide. If you tell the agent that you want to watch Star Wars, it will figure out whether you subscribe to the streaming service, and if not, it will proactively sign up for you. If you really don’t know what you want to watch, it will actively provide suggestions and play the content you specify.

You also get news and entertainment content tailored to your interests. CurioAI supports custom podcasts (based on any topic you’re interested in), which can also provide a glimpse of what’s to come.

Impact on the technology industry

In short, intelligence will be helpful in almost any field of life and human activities, and it will also have an impact on software. have a profound impact on the industry and society.

In the computing industry, we talk about platforms—the technologies on which apps and services are built. Android, iOS, and Windows are all platforms. Agents will become the next platform.

You don’t need to know how to write code or graphic design, just tell the agent what it wants and you can create a new application or service. The agent automatically writes code, designs application skins, logos and publishes programs to online stores. We see a future with OpenAI launching GPT this week, making it easy for non-developers to create and share their own assistants.

Intelligent agents will also affect the way we use and write software. They will replace search sites because they will do a better job of finding and summarizing information. They will also replace many e-commerce sites because they will not be bound by the number of suppliers to automatically find the best prices. They will replace word processors, spreadsheets, and other productivity applications. Today's separate businesses—search advertising, ad-supported social networks, shopping, productivity software—will become one.

I don’t think the intelligent agent business will be dominated by any one company. There will be many different AI engines in the future. Today, agents are embedded in other software (such as word processors and spreadsheets), but eventually they will run autonomously. While some agents are free (with advertising sponsorship), I think most agents are paid so that companies have an incentive to make the agents work for users (rather than advertisers). Extremely fierce competition will make agents very cheap.

However, before what I describe can become a reality, we still need to face some problems. Questions about this technology and how we will use it. I have written before about problems caused by AI, and now I will focus mainly on problems caused by agents.

Technical Challenges

No one has yet figured out what the agent’s data structure will look like. To create personal agents, we need a new kind of database that can capture all the nuances of your interests and relationships and quickly recall information while preserving privacy. We've seen new ways of storing information, such as vector databases, that may be better suited for storing data generated by machine learning models.

Another open question is how many agents people will interact with. Will your personal agent be separate from the agent of your therapist or math tutor? If so, when do you want these agents to work together, and when should they be left alone?

How will you interact with the private agent? Companies today are exploring different strategies, including apps, glasses, pendants, pins, and even holograms. All of these are possible, but I think the first big breakthrough in human-computer interaction will be wireless earbuds. If the agent needs to contact you, it will talk to you or display on your phone. (For example, your flight is delayed. Do you want to wait, or do you need me to rebook?) If you choose, it will also monitor the sound entering your ears, by blocking out background noise, amplifying hard-to-hear sounds, or Enhance your voice by making it easier for you to understand people with strong accents.

There are other challenges. There is currently no standard protocol that allows agents to communicate with each other. In order for everyone to be able to use it, the cost of the intelligence needs to be reduced. The method of prompting the agent to give the correct answer is also simpler. There is a need to prevent hallucinations, especially in areas where accuracy is critical (such as health), and to ensure that agents do not harm humans because of their biases. Hopefully the agent won't do anything it shouldn't do. (This refers to humans using intelligence for evil purposes).

Privacy and other important issues

Given all these challenges, online privacy and security issues will also become more pressing. You want to be able to decide what information an agent can access, so you can be confident that your data is only shared with selected people and companies.

But who owns the data you share with the agent? How to ensure that this data is used appropriately? No one wants to see an ad related to something they told their therapist. Can your agent be used by law enforcement as evidence against you? When does an agent refuse to do something that might be harmful to you or others? Who picks the values ​​that the agent needs to align with?

There is another question, how much information should the agent share. You want to meet a friend and let the agent do the communication. You don't want it to say, "Oh, she's meeting other friends on Tuesday and doesn't want to include you." If you ask the agent to write a work email for you, it needs to be told no. Private information about you or proprietary data from previous employment should be used.

Many of these issues have become top priorities for the tech industry and lawmakers. Recently, I participated with other technology leaders in an artificial intelligence forum organized by Senator Chuck Schumer and attended by many U.S. senators. We shared our thoughts on these and other issues and discussed the need for lawmakers to pass strong legislation.

But other issues will not be decided by companies and governments. For example, agents may influence how we interact with friends and family. Today, you can show someone that you care by remembering details of their life—such as birthdays. But does it make just as much sense to them when they know your agent will remind you and deliver the flowers?

In the distant future, intelligent agents may even force humans to confront profound questions about purpose. Imagine they become so good that everyone can have a high quality of life without having to work too much, what will people do with their time? When the agent knows all the answers, there will still be people who want to accept the rewrite: Education is a very important field that has a profound impact on the development of individuals and society. The purpose of education is to develop students' knowledge, skills and values ​​to enable them to succeed in future life and work. The education system should provide fair opportunities and ensure that every student receives a high-quality education. In addition, education should also focus on cultivating students' creativity, critical thinking and problem-solving skills to cope with increasingly complex social challenges. To achieve these goals, educational policies and teaching methods need to be constantly innovated and improved. Education is not only the responsibility of schools, families and society should also actively participate in and support the development of education. By working together, can we build a more prosperous and progressive society? Can there be a safe and prosperous society when most people have a lot of free time?

However, we are still a long way from such a future. At the same time, intelligent agents are coming to us and will completely change human online and offline lifestyles in the next few years.

Reference link
https://www.gatesnotes.com/AI-agents

The above is the detailed content of Bill Gates personally revealed the importance of “OpenAI Spring Festival Gala”. For more information, please follow other related articles on the PHP Chinese website!