Source: Titan Media

Image Source: Generated by Unbounded AI

The new wave of AI revolution is profoundly impacting human society. On the path to the next generation of General Artificial Intelligence (AGI), in the foreseeable future, in a world of deep coexistence between humans and machines, what revolutionary disruptions will occur in our current social behaviors and rules?

On the morning of December 1st, Beijing time, at the T-EDGE Global Innovation Conference 2023 in California, Sebastian Thrun, the father of Google's self-driving car, founder of Google X Lab, and CEO of Sage AI Labs, issued a brand-new and weighty warning for the first time regarding the revolution sparked by the new wave of artificial intelligence.

In his keynote speech titled "The New Era of Artificial Intelligence," Thrun believes that AGI is an intelligent version comparable to human intelligence, and in some fields, we have already achieved AGI. However, what we have not yet possessed is an artificial intelligence capable of performing all these tasks.

Based on over 20 years of deep cultivation in the field of artificial intelligence, Thrun stated that within the next three years, the following five major areas of work that rely on human intelligence will be disrupted. We will see AIGC being on par with human intelligence workers in all fields, which means that all jobs will undergo changes, and no job will remain unaffected.

Medical Industry

AI systems for diagnosis in many fields such as radiology and skin cancer can detect potential deadly diseases long before people notice symptoms; AI can provide personalized precision medicine tailored to your background, DNA, and personal characteristics, aiming to produce the maximum impact with minimal harm, abandoning the era of "one-size-fits-all"; AI also plays a significant role in drug development.

Legal Industry

AGI is already able to pass the California Bar Exam and is no less competent than graduates from any leading law school. Tasks related to discovery, attempting to find necessary facts, conducting research, and drafting documents in legal work can now be handled by artificial intelligence.

Automotive Industry

The cars we have now can drive more confidently and safely, outperforming human drivers.

Customer Service Industry

AI tools in customer service centers can extract the best behaviors of top customer agents, leading to a 14% increase in productivity.

Education Industry

AI tutors can provide personalized guidance and truly understand each learner, and they can respond immediately day and night.

This also means that we will redefine the behaviors and operations of society. Some AIs are on our side, they can represent us and serve us, and we can trust them, which will be the greatest disruption to society in history.

Below is the transcript of Sebastian Thrun's speech at T-EDGE 2023, edited by the Titanium Media app:

Hello, everyone. My name is Sebastian. Welcome to today's T-EDGE conference held in Beijing. I'm very sorry that I couldn't attend in person, but I'm still very happy to meet everyone in this way.

Let's talk about 2017, when Google's Google Brain team released a paper titled "Attention is All You Need," which applied the same machine learning concept to a broader architecture, effectively integrating dispersed databases and machine learning, sparking a revolution.

Today, when we see modern systems like GPT-4 taking over and effectively performing at a human level, we can all feel this revolution. This is the architecture, and I won't spend time explaining it because it's complex and has many layers, but I suggest delving into it and fully understanding it, as it's an excellent work that can apply machine learning to a large number of new systems, including in this case, dictionaries and databases.

Now we have all experienced recent events, such as the emergence of GPT, as a new way to interact with computers. In many ways, GPT represents a progressive understanding of technology, an evolution in understanding what this technology can do.

The question today is, is GPT an "black swan" event in the field of artificial intelligence?

Let me explain what a "black swan" event is. This term was coined because black swans are rare, and discovering one signifies a huge surprise. In the field of computers, I have experienced two black swan events in my lifetime, one being the emergence of the internet, and the other being the invention of the iPhone.

The internet first allowed people to communicate across oceans and many countries in a borderless digital world, and the emergence of the iPhone further put the internet in your pocket, creating new businesses such as Airbnb, Uber, and China's Didi.

Therefore, it is entirely imaginable that in the next 20 years, we will see new companies rise and new business models emerge, which will be completely great for humanity, and there are a large number of applications being pursued today.

Clearly, when you see transformers in large language models, you will find that they answer questions in unprecedented ways. So, if your business involves customer interaction and customers conducting research and asking questions, you can now redefine the way this interaction occurs, summarizing a large amount of information and presenting it in a form we understand.

Why should a book have 300 pages? Why can't you condense this book to 10 pages with the press of a button? You can read it in 10 minutes. In terms of drafting documents, you have seen the unparalleled ability of GBT in drafting documents. If you haven't seen it yet, visit GPT or any other toolbox and ask it to draft legal documents, a song, or a love poem—whatever you need, it can draft it for you.

In terms of machine translation, GPT can translate into 90 different languages, and the level of speech recognition is very good now, almost comparable to a native speaker, which means you can communicate in any language you want. I can fluently converse with our Chinese nanny in Chinese, but when I talk to my phone, I use English. There are now excellent oral and written Chinese writing software.

We believe this is the pinnacle of intelligence in software engineering. You can now code in any language, and you have a toolbox that can speed up your code by 10 times, and even we think uniquely human things, such as creating art.

I will give an example later to show how artificial intelligence competes with human performance in my work applying these technologies. One of the primary ways we utilize artificial intelligence is through a company called Cresta, which I have been running with my students for 5 years.

At Cresta, we provide AI tools for customer service centers, where people work in call centers and customer service centers, interacting with customers, selling products to customers, handling customer concerns, and providing customer service and retention. Now, as every call center operator knows, some calls are excellent for business, while others are less ideal, depending on their abilities, skill levels, and education.

Through artificial intelligence, they can extract the best behaviors of top customer agents, and every time a customer agent satisfies a customer, we have a positive training example. Then, we train the artificial intelligence brain, this large language model, to understand the elements that make up excellent customer interactions, and then empower everyone with this training signal.

Massachusetts Institute of Technology and Stanford University have just released a large-scale study in collaboration with partners in the field. We will immediately see a 14% increase in productivity. Although this number seems small, I am quite disappointed with it, considering that the United States spends billions of dollars annually in call centers and contact centers, this is still a huge amount of money.

I also use large language models for education. The pain point in education is getting personalized guidance, having someone who understands you, and providing personalized training at your pace and level of knowledge to ensure that you and your children can fully develop in the field of education. In my global tech education company Udacity, we transitioned from human tutors to AI tutors this spring.

Although AI tutors may not be as excellent as human tutors, they are still much better and excel in certain aspects, as they truly understand each learner and can respond immediately day and night. On the day we introduced AI tutors based on large language models, our interactions and engagements increased from 200 times a day to 20,000 times a day. How much of a difference does this make in turning AI systems into personalized tutors?

This is just the beginning. I dream of a world where every child, every person in the world can unleash their full potential because they have an AI tutor that can be established almost without cost, enabling them to have the ability to learn anything they want.

Therefore, their only limitation will no longer be where we live, where we grow up, or how much money we can spend on education, but rather, we can all participate in the best education globally, regardless of age, background, or geographical location.

This will be a game-changing revolution. There are still many other aspects where artificial intelligence will change human health. I have already mentioned today the diagnostic AI systems in many fields such as radiology and skin cancer, which can detect potential deadly diseases long before people notice symptoms. Let us remind everyone that cancer is a treatable and curable disease, as long as you can detect it early.

Furthermore, you will see the era of personalized precision medicine, where drugs are tailored to your background, your DNA, and your personal characteristics to produce the maximum impact with minimal harm, abandoning the era of "one-size-fits-all" and entering the era of personalized precision medicine, which will have a profound impact on medicine.

Finally, AI plays a significant role in drug development, where what used to take weeks or even months of work now only takes a millisecond in critical aspects of drug design. In this process, many reliable companies have emerged in the United States and China, let us seriously pursue the holy grail of medicine, methods to slow down aging and cure it. Imagine if a method to stop or slow down aging is found, giving people a longer and better life. You would then have the biggest business opportunity in history, wouldn't you?

Now, artificial intelligence will impact many other professions, one of which is the legal profession, which is very important because at least in the United States, legal fees are very expensive. I pay about $1000 per hour for my best lawyers. That's a lot of money, equivalent to the monthly income of others.

So how will it impact the legal profession? A large part of the work of lawyers is related to discovery, attempting to find necessary facts, conducting research, and drafting documents, all of which can now be very competently handled by artificial intelligence. Many AI companies are entering this field and providing related services, but as it develops, artificial intelligence may even do more, it may become your lawyer, your judge when resolving conflicts.

Today, almost every human interaction is constrained by some legal background, whether it's a contract, a law, or just a convention, and almost all of these interactions lack a deep understanding of someone who knows the law. In the future, the law may become an all-encompassing subject that affects all of us every day, making us better through better compliance and a deeper understanding of operating rules, but the ultimate goal will be AGI, which will change all of us.

What is AGI? AGI stands for Artificial General Intelligence, it is an intelligent version comparable to human intelligence. I believe we have already achieved AGI in some fields. The cars we have now can drive more confidently and safely, outperforming human drivers. These artificial intelligence systems have already performed as well as graduates from any leading law school in California's bar exam, and these graduates have become lawyers in California.

But what we have not yet possessed is an artificial intelligence capable of performing all these tasks. I believe we are very close to it, and within about three years, I believe you will see artificial intelligence systems being on par with human intelligence workers in all fields, which means that all jobs will undergo changes, and no job will remain unaffected. This means that we will redefine the behaviors and operations of society. With artificial intelligence systems on our side, they can represent us and serve us, and we can trust them, which will be the greatest disruption to society in history.

It all stems from a very simple idea: Can computers be trained? Can they learn?

The answer is yes. This can be shown by giving two examples, one of which comes from a recent paper, similar to the calculation of intelligent systems, showing a logarithmic graph, where in all three cases, the horizontal and vertical lines are represented logarithmically, measuring complexity, such as the amount of computation used to train networks, and the size of the data input into the network is usually hundreds of billions of documents. The number of parameters found in the network is a measure of testing rules, and through testing rules, we measure the number of errors these networks make.

One such test covers English language tests, legal tests, or driving tests for self-driving cars represented logarithmically. Historically, all these systems have shown as a straight line. Why is this important? Because over time, our systems will grow exponentially, thanks to Moore's Law. Therefore, the horizontal logarithmic form is basically a scale of time, and it also means that over time, our losses and error rates will exponentially decrease, they become smaller and smaller.

So we live in a world where we see exponential progress over time, thanks to this recorded phenomenon, decision-making becomes better and better, and we don't see it stopping from becoming smarter, it's just a matter of time, they will be smarter than all of us.

Including myself, once we have artificial intelligence companions, we will increase our efficiency by 100 times in what we do today, which means that what you and I do in two weeks can be completed in 1 hour. Isn't that amazing? This will happen, and I hope in my lifetime, which means it will provide more innovation opportunities for all of us, more creativity, and discovering more things we are eager to start a new life.

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。