Shulou(Shulou.com)12/24 Report--

[new Zhiyuan Guide] on the first anniversary of the release of ChatGPT, Sam Altman restarted OpenAI. Let's start with the end of Game of Thrones and review OpenAI's "ChatGPT year".

On the occasion of the first anniversary of ChatGPT, OpenAI announced that Sam Altman restarts OpenAI!

A year ago today, as CEO, he was doing the final preparations for the launch of ChatGPT. He should never have imagined that so many things would happen in the year after going online.

Today, OpenAI also put on a party hat to celebrate his birthday.

Put the mouse over the hat, there will be such a line: "this is my first birthday!" "

Altman pulled back OpenAI, who had been standing on the edge of the cliff, and set a new direction.

In an open letter to all OpenAI employees, he clearly set three short-term goals for the company:

First, we should make every effort to promote the research work in accordance with the plan and further strengthen the full-stack security research.

Second, continue to develop and deploy products to serve customers well. So that more people can benefit from AI and let human beings adapt to AI.

Third, the three founding board members will promote the reform of the corporate governance structure and ensure the stable operation of the company in the future.

At the same time, the company's senior management team officially laid off staff.

Greg Brockman is still chairman of OpenAI, and Altman continues to serve as CEO,Mira as CTO, responsible for the day-to-day operations of the company.

The position of Ilya, the company's former chief scientist, has not yet been settled, according to Altman in the open letter: "We will continue to discuss with Ilya the way he works with you."

The three members of the newly established founding board of directors are also ready, namely:

Chairman: Bret Talyor

Larry Summers

Adam D'Angelo

As chairman of the board of directors, Bret Talyor also wrote an open letter to employees on behalf of the founding board of directors.

In his letter, he also promised to promote three tasks:

First, establish a diversified and competent board, and Microsoft will get an observation seat without the right to vote.

Second, the board of directors should conduct an independent investigation into the recent palace fighting.

Finally, the board of directors should upgrade the governance structure of the whole company to ensure the stability and prosperity of OpenAI in the future.

Now, OpenAI's "Game of Thrones" drama has finally come to an end.

After more than 10 days of ups and downs, OpenAI finally completely reshuffled the board of directors of the "six King Kong", with only Adam D'Angelo remaining in office and all the other five people leaving.

Microsoft finally won a non-voting seat and was able to avoid the fate of "spending 13 billion but being the last to be notified."

The end result of the company's executive team is that Ilya leaves and the other three stay on.

The latest interview with Altman

Altman said in an interview with foreign media: it is clear that there are problems with the governance structure of OpenAI.

As for whether the non-profit governance structure, which is very curious to the outside world, will change, Altman replied that it still needs to be considered by the board of directors in the long term, and there will be no answer in the short term. Please be patient.

He still declined to say why he was fired in the first place, but the board will seriously investigate the issue.

However, he said he was so upset when he learned that he had been fired that it was hard to imagine that he would leave the company where he had been working for four and a half years.

When the media asked him if he had learned anything from the whole palace struggle, he replied: I have learned a lot. I can't finish talking at 01:30, so I won't talk about it for a while.

When asked if he wanted to go back to the board, Altman said he didn't have time to focus on the issue and had too much to do.

Altman also declined to comment on the leaked QQ Magna Altman.

ChatGPT this year November 30, 2023, is the first anniversary of the birth of ChatGPT.

I don't know if you're a little surprised.

Amazed at the passage of time and the changes around us.

A year ago, we didn't have ChatGPT.

There is no concise interface to enter "prompt words" and get a detailed reply in a few seconds.

Will not discuss "the new artificial intelligence" with family, friends and colleagues.

The artificial intelligence that children use to do their homework, the artificial intelligence that can "hallucinate"

The artificial intelligence that dominates all the headlines, the artificial intelligence that will change or destroy the world.

A year has passed, people come and go, flowers bloom and leaves fall, and the AI wave led by ChatGPT has really profoundly rewritten the world we live in.

Next, let's follow the pace of time and look back on the 12 months that ChatGPT changed the world.

Looking back on its debut, it was questioned.

Reviewing its amazing ability in the big model competition, it has become the benchmark of the industry.

Review its evolution to GPT-4 Turbo at the developer conference.

Of course, it is also necessary to make the whole network of the six King Kong, as well as the love and hatred between Altman and the board of directors.

November 2022: ChatGPT debut on November 30, 2022, at the NeurIPS machine learning conference in New Orleans, rumors were flying about GPT-4, and it was speculated that details of GPT-4 would be revealed at the conference.

However, OpenAI released text-davinci-003, known as part of the "GPT-3.5 series".

It is improved on the basis of previous models by dealing with more complex instructions and generating higher quality and longer forms of content.

At the same time, OpenAI launched the so-called ChatGPT "early demonstration" almost after the fact.

ChatGPT adopts "interactive dialogue mode" and is able to answer follow-up questions, admit mistakes, question incorrect premises, and reject inappropriate requests.

Soon, ChatGPT became the fastest-growing consumer application in history.

December 2022: two weeks after the release of ChatGPT, everyone is talking about it. However, while there was a lot of hype, there were a lot of criticisms.

From "sensational" to "the most subversive technology" to "chattering and eloquent, both facts and nonsense".

OpenAI CEO Sam Altman tweeted, "ChatGPT's function is very limited, but its performance in some ways is enough to make people think it is excellent, this is a rehearsal for progress." We still have a lot of work to do in terms of robustness and authenticity. "

January 2023: ChatGPT will be disabled at the top of AI in January this year, causing a heated discussion about how enterprises, organizations, and institutions will respond to the rise of large language models.

LLM can help human beings-whether the answer is borrowed, expanded, or plagiarized depends on the human point of view.

When soliciting papers, the International Machine Learning Congress pointed out that "papers are prohibited from containing texts generated by large-scale language models (such as ChatGPT) unless the generated text is displayed as part of the experimental analysis of the paper." "

February 2023: the competition between competitors and ChatGPT is in full swing by February, when Google announced a $300 million investment in Claude's Anthropic, one of OpenAI's hottest competitors.

Google will hold about 10 per cent of the shares, valuing Anthropic at about $5 billion, according to the Financial Times.

Anthropic was founded by several researchers who left OpenAI in 2021.

In April 2022, less than a year after its establishment, Anthropic suddenly announced that it had raised up to $580 million in financing, thus gaining public attention.

It turned out that most of the money came from employees of Sam Bankman-Fried and FTX, who founded the cryptocurrency platform FTX (FTX is accused of fraud and now bankrupt and Bankman-Fried has been arrested).

March 2023: just four months after GPT-4 's blockbuster release of ChatGPT, OpenAI unexpectedly released its long-awaited GPT-4 model.

OpenAI called GPT-4 "the most advanced system that can make safer and more useful responses".

As an upgraded version of ChatGPT, GPT-4 can solve more difficult problems more accurately with a wider range of common sense and problem-solving skills.

It also adds new features, such as accepting pictures as input and generating titles, categories, and analysis.

GPT-4 can handle more than 25000 words of text and can be used to create long content, expand conversations, and search and analyze documents.

April 2023: Sam Altman travels around the world. It was a beautiful spring for OpenAI, when Sam Altman held a face-to-face meeting with the Japanese Prime Minister in April, launching a global spring trip.

A total of 17 cities were visited to promote OpenAI,Altman.

May 2023: plug-in features launch OpenAI announces the launch of the ChatGPT plug-in to ChatGPT Plus subscribers.

The company said the beta version allows ChatGPT "to access the Internet and use more than 70 third-party plug-ins." "

These plug-ins include functions for playing chess, finding recipes, live football and simple nutrition.

June 2023: Google Gemini surfaced in the summer, and competition for LLM is heating up again. Google brain and DeepMind teamed up to form Google DeepMind.

Demis Hassabis, CEO of DeepMind, said the company was developing a new system, Gemini.

Gemini will combine LLM technology with reinforcement learning technology used in AlphaGo, with the goal of giving it new ability to plan and solve problems.

July 2023: the GPT-4 model that denies GPT-4 's "laziness, stupidity" that OpenAI supports ChatGPT has become "lazier and stupider" as a result of a "complete redesign".

OpenAI's response is that "as you use it more frequently, you will begin to notice problems you haven't seen before." "

August 2023: it was expected that OpenAI launched an enterprise version of OpenAI for enterprise customers, but it was a bit of a surprise when it launched ChatGPT for enterprises in August.

September 2023: ChatGPT revisits the Internet one of ChatGPT's biggest weaknesses from the beginning is its knowledge limitations-it contains only information as of September 2021.

But in September, OpenAI announced that ChatGPT "can now browse the Internet, provide users with up-to-date authoritative information and provide direct links"

This is due to the integration with Microsoft's Bing search engine.

Previously, networking was removed because it skipped the paywall.

October 2023: DALL ·E 3 introduces ChatGPTOpenAI to users of ChatGPT Plus subscription services and ChatGPT for Enterprise, introducing the latest image generation AI model DALL ·E 3.

DALL E 3 reliably renders complex details, including text, hands, and faces. In addition, it is particularly good at responding to a wide range of detailed prompts.

November 2023: the developer conference and ChatGPT11 month of "Game of Thrones" once again ushered in the best part. OpenAI's first developer conference unveiled a series of products-- customized GPT! New GPT-4 Turbo! Assistant API.

The wave of user excitement has swept Silicon Valley and the world.

Then there is the drama of "OpenAI Power Tour", which just ended, pushing the legendary year of ChatGPT to the best part!

Reference:

Https://venturebeat.com/ai/chatgpts-1-year-anniversary-how-it-changed-the-world/

Https://www.theverge.com/2023/11/29/23982046/sam-altman-interview-openai-ceo-rehired

Https://openai.com/blog/sam-altman-returns-as-ceo-openai-has-a-new-initial-board

Https://twitter.com/OpenAI/status/1730030975931846939

This article comes from the official account of Wechat: Xin Zhiyuan (ID:AI_era)

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.