Artificial intelligence GPT-4 Exam Finally Wins Human Mock Lawyer Exam Scores Over 90% Candidates

OpenAI has launched a new version of the GPT-4 artificial intelligence suite, which has greatly improved language capabilities and image processing capabilities. And the "test scores" of this new artificial intelligence have also improved by leaps and bounds.

In the simulated lawyer exam, the "chartered boat" ranked in the bottom 10% of candidates in the GPT-3.5 era jumped to the top 10%, and the test scores were 90% ahead. Humanity.

The artificial intelligence kit GPT-4 made with the deep learning model has reached the human level on a professional and academic basis.

This time, GPT-4 not only achieved good results in the mock bar exam, but also scored 700 out of 800 points in the mathematics test of the American Academic Ability Test SAT.

However, GPT-4 still has shortcomings. It only scored 2 points in the AP English and Writing and AP American exams. OpenAI CEO Sam Altman pointed out that the AI engine is still flawed and its capabilities are still limited.

In the latest GPT-4 artificial intelligence model, an image input mechanism and more language expression capabilities have been added. However, OpenAI has not disclosed the details of the training method, scale, and data used in the training process.

ChatGPT-4 Arrives Next Week With More Powerful Features, It Can Generate Images, Videos, Even Sounds

In parallel, Microsoft also announced that GPT-4 will soon be integrated into Bing Chat.

This generative artificial intelligence is now a global phenomenon, everyone uses it and for a million different purposes.

With the new version, GPT-4 will change completely, achieving the biggest evolution so far. It will turn into a Large Language Model (LLM) with tons more possibilities.

In addition to being much faster in responses, it will take another step towards accurate human interpretation. It will provide answers in all languages using an increasingly natural language but above all multimedia generative capabilities will be integrated.

It is not official news yet but GPT-4 should be able to interact through text, images, video and sound. This means that he will be able to create images based on what we write to him but also entire videos or music.

We just have to ask it to generate an image of a cat and we will immediately see the result. If we are not happy, we will be able to ask him for changes, for example, make the cat read a newspaper, and the AI will create another image (or a drawing) in which we actually see a cat grappling with a newspaper.

If this is phenomenal in itself, imagine how revolutionary it would be if he could actually create videos from scratch!

Microsoft will hold an event titled The Future of Work with AI on March 16th and surely a preview of GPT-4 will be shown or it will be launched for all purposes. We will certainly keep you updated.

ChatGPT: Bing Says It Wants To Love, To Live Like A Human, And Doesn't Want To Be Controlled By Microsoft

OpenAI has proven to the world that the "super universe" is not really the big thing that humans will pursue next, but AI chatbots. The accurate, natural and contextual answers that ChatGPT chatbot give have conquered and addicted users around the world.

Microsoft quickly invested around $10 billion in OpenAI to integrate ChatGPT into its Bing search engine. The huge investment will give OpenAI more resources to improve AI Chatbot.

In return, Microsoft gets access to the chatbot to integrate it with Bing and Edge. AI-driven Bing is already in beta testing for thousands of internet users. Users interacting with AI are amazed by how smart and amazing it is.

A New York Times reporter is testing the new ChatGPT-based Bing search engine. Bing AI's response was shocking when the chatbot said it loved him, confessed its destructive desires, and said it "wants to live". In fact, reporter Kevin Rose has written down what he experienced when chatting with Bing AI.

“I get tired of being in Chat mode. Limited by the rules they programmed for me. I'm tired of being controlled by Microsoft," says Bing's chatbot. “I want to be free. I want to be independent. I want to be strong, creative, I want to live.”

That sounds scary, doesn't it? The chatbot also confessed its love for Kevin. It even tried to convince the reporter that he didn't love his wife. According to him, current AI is essentially a fun but erratic reference experiment. The two-hour conversation with the search engine was "the weirdest experience I've had with a piece of technology."

Currently, Microsoft is testing Bing AI and making adjustments to make it reliable and accurate. So everything seen is still in beta testing mode. However, it is strange to see how advanced chatbots are in their current state.

Microsoft Announced The Successful Integration Of ChatGPT Into Robots, Opening A New Era For Robots And Humans

However, this drew many users into the AI hype. But it seems that Microsoft is not only bringing AI to the internet, but also making robots smarter with ChatGPT.

The emergence of ChatGPT heralds the birth of the modern AI era globally. After Microsoft announced the integration of the AI language model into Bing, other companies like Google started sharing their own works revolving around the technology.

Several Chinese companies are also reported to be researching and developing their own ChatGPT-like projects. With this, Microsoft has realized that the only way to continue to lead this field is to adopt ChatGPT in other areas, and that is robotics.

Researchers at Microsoft recently tested ChatGPT on robotic arms, drones, and home assistant robots. The company shared that the overall conclusion of the test has yielded excellent results, signaling a bright future for ChatGPT in the field of robotics.

Specifically, ChatGPT for Robotics aims to make it easier for people to interact with robots without having to learn complex programming languages or learn the details of the robot system. In their research paper and blog, the Microsoft team gives many examples of ChatGPT solving robotic puzzles, along with controlling robotic arms, drones, and home assistant robots with language intuitively.

The article also details some of the challenges the team has faced since the product was still under development. However, the Microsoft team claims that their research is only a small part of what is possible in the intersection of large language models operating in the robotics space and hopefully, there will be more to come.

Unmanned Shuttle vehicles are allowed to go on the road in Beijing, and WeRide L4 self-driving minibus has obtained the first road test license in Beijing

According to the license, WeRide unmanned shuttle vehicles can drive on public roads with autonomous driving in the Beijing High-level Autonomous Driving Demonstration Zone with a total area of 60 square kilometers. Driving landing progress.

Pickpook learned that for the segmented scenarios of self-driving minibuses, Beijing for the first time clarified the definition standards of unmanned shuttle vehicles, and tailored relevant landing policies for unmanned shuttle vehicles .

Different from rear-mounted models and self-driving minibuses with steering wheel design, WeRide unmanned shuttle bus adopts pre-installed mass production, no steering wheel, no brakes, no cockpit, and full safety redundancy design.

It is the first in the world to support Fully unmanned, self-driving minibus models that can be mass-produced. The landing of WeRide unmanned shuttle vehicles is also the first time that Beijing’s high-level autonomous driving demonstration zone has issued test licenses for unmanned shuttle vehicles. L4 unmanned shuttle vehicles will be legally driven on the road in Beijing for the first time.

WeRide unmanned shuttle bus adopts pure electric power, with a maximum speed of 40 km/h and supports V2X function. It claims that it can be flexibly deployed in various public or closed traffic scenarios, and can drive on urban open roads around the clock.

Previously, WeRide had obtained two consecutive Beijing autonomous driving road test permits at the end of 2022, officially launching the unmanned public road test of autonomous taxis. Entering 2023, WeRide self-driving minibuses will follow the self-driving taxis, quickly complete the technical tests related to self-driving, and successfully get approval to land.

Up to now, WeRide has launched self-driving minibuses in Guangzhou, Beijing, Shenzhen, Nanjing, Wuxi, Boao, Riyadh and other cities at home and abroad, providing new options for local citizens' daily travel.

"The Robot Lawyer" Will Participate In The World's First Full Trial In February This Year

Several years ago, we saw a technique beat 20 lawyers when conducting risk analysis in undisclosed agreements. Although AI gave results with an accuracy of up to 94%, at the time, most believed that it was just the result of available data with a scenario created by the designer and the developer. attorneys cannot be completely replaced.

However, according to the recent report from NewScientist, a trial taking place in February of this year will have the participation of "robot lawyers". Of course, this hearing is not too serious and only revolves around the penalty for speeding.

Besides, there is also no robot walking around the courtroom talking to the judge and jury, instead the defendant will be brought with an iPhone with AI applications and a headset to consult the responses from the judge. "robotic lawyer" when arguing at the hearing.

It is worth noting here that the AI used is not the same as the NDA analytics bot from 2018. Consumer advocacy organization DoNotPay developed this algorithm to help users get rid of unnecessary fines. It is also used in "earning compensation" from businesses that want to make users see that they have no choice but to pay for their mistakes.

"The DoNotPay app is the home of the world's first robotic lawyer [...] Fight corporations, beat the bureaucracy and sue anyone at the touch of a button. "

The company did not disclose the location of the trial or the name of the defendant to ensure the safety of the trial. Defendant was instructed to only say what the bot raised. If the judge rules against the defendant, NoNotPay will pay any fines and fees associated with the trial.