Microsoft CEO Satya Nadella came on stage for OpenAI CEO Sam Altman’s keynote speech to talk about the future of their partnership.
Tiernan Ray
On Monday in a downtown event space in San Francisco, OpenAI, the creator of ChatGPT and GPT-4, held its first conference for developers, Dev Day. The company unveiled GPT, a way to easily bring together custom versions of ChatGPT, a store to find and purchase custom ChatGPT, to make it easier for developers to call specific functions for their applications. A “Helper” API, and many other features and upgrades. ,
The only swag so far is the OpenAI logo and a bunch of nifty pins labeled like “Engineering” and “Research” and “Go to Market.” Another set of pins represent personal pronouns.
Tiernan Ray Tiernan Ray
First, Sam Altman took the stage and reiterated various milestones: ChatGPT a year earlier, followed by GPT-4, which is “still the most powerful model”.
The company revealed that it has more than 2 million developers using its API “for a wide range of use cases” and 92% of Fortune 500 companies. ChatGate itself gets about 100 million weekly active users, the company said.
Also: MacBook Pro (M3 Max) review: A desktop-class laptop for the AI-powered era
Altman then began rapidly listing the many innovations being announced. There was thunderous applause at various announcements – it’s an excited, enthusiastic crowd.
Altman brought a special guest to the stage: Microsoft CEO Satya Nadella. Altman jokingly asked Nadella, “What is Microsoft thinking about a partnership?” Both Nadella and the audience laughed a lot at this.
“You guys have created something magical,” Nadella said. Nadella said the partnership has “dramatically changed” the shape of Microsoft’s Azure cloud computing service. “Our job is to build the best systems so you can build the best models,” Nadella said.
“The first thing we’re doing in partnership with you is building the system,” Nadella said. “We want to be your co-pilot as developers on the OpenAI API.”
“Some things will be very important to us,” Nadella said. “We are absolutely committed to ensuring that you not only have the best models but also the best computations,” Nadella said. “Our mission is to empower every person.”
“I always think of Microsoft as a platform company, a developer company, and a partner company,” Nadella said. “We have to be at the top of our game to have the systems that are required as you move forward aggressively on your roadmap,” Nadella said. “The shared mission of both companies is to empower every person in every organization on the planet to achieve more,” Nadella said.
In response to Nadella, Altman commented, “I’m excited to build AGI together,” meaning artificial general intelligence, the notion of computers that can match human thought abilities.
Dev Divas venue, SVN West event venue at Van Ness and Market in Downtown SF.
Tiernan Ray
Major product and technology news include:
- GPTs: Custom versions of ChatGPT that OpenAI says “anyone can easily create” for specific tasks. The company is offering two initial custom GPTs for popular design apps and workflow software, Canva and Zapier AI, respectively. The company plans to introduce additional GPTs;
- GPT Store: Later in November, OpenAI will open a GPT Store to receive GPTs created by others, where developers can earn money for their creations;
- Copyright Shield that will cover the costs of protecting customers;
- fine-tuning service for GPT-4 for developers;
- Custom Models Program for Enterprises, a team of OpenAI researchers that will work with “selected organizations” to “train custom GPT-4s in their specific domains”;
- A new user interface for ChatGPT, featuring a simple, dark background with the OpenAI logo and the phrase, “How can I help you today?” The new user interface will make it easier to coordinate ChatGPT and OpenAI’s image-generation program DALL-E, the company said;
- GPT-4’s information source is updated until April 2023, which is a big step up from the traditional limit imposed on the program in September 2021. ChatGPT also gains the ability to search PDF and other documents;
- The GPT-4 program gets its own “context window”, quadrupling the amount of input it can take when generating answers from 32,000 to 128,000 in the new “turbo” version of the program. (For more information about the various features of the GPT model, see OpenAI’s web site.);
- GPT-4 Turbo can now accept images as part of prompts, and produce “human-quality speech” as its output;
- Assistant API: A function-calling mechanism that makes it easy for developers to plug specific “assistant” functions into their apps, such as “a natural language-based data analysis app, a coding assistant, an AI-powered vacation planner, a voice – Controlled DJ, a smart visual canvas.”
- A new “seed” parameter makes GPT return “reproducible output” “most of the time”;
- A new version of GPT-3.5 Turbo that gets improved function handling and JSON handling;
- Cut the price of GPT-4 Turbo and GPT-3.5 Turbo based on the price per input and output token, and double the “tokens per minute” rate that can be used.
Altman demonstrated GPT on stage, writing from the ground up a program called Startup Mentor to mentor entrepreneurs. He uploaded a file of a conversation he gave, as an example of input through external sources. The program is designed to answer questions like, “What are three things startups should look for when hiring people?”
“We won’t be able to do this with many companies in the beginning, and it won’t be cheap,” said Altman of the custom model.
The Copyright Shield program, Altman said, “means that we will take steps to protect our customers” in the event of litigation, “and cover the costs.”
The Copyright Shield program, Altman said, “means that we will take steps to protect our customers” in the event of litigation, “and cover the costs.”
Tiernan Ray
Also: Overseeing generic AI: New software leadership roles emerge
OpenAI says a key element of the Assistant API is “persistent threads”, which “allow developers to avoid resending the entire conversation history with every new message and work around context window constraints.”
Tiernan Ray
The new GPT-4 Turbo can be accessed immediately in preview form, OpenAI said, bypassing the command gpt-4-1106-preview for the OpenAI API. A stable version is planned to be released “in the coming weeks”.
GPT-4 Turbo works better when following specific instructions, such as outputting responses in XML, the company said. It also gains the ability to have responses in JSON via a new parameter, “response_format”.
Also: Robots Plus Generative AI: Everything You Need to Know When They Work as One
The new seed parameter for GPT-4 is a beta feature that is “useful for use cases like re-running requests for debugging, writing more extensive unit tests, and generally having a higher level of control over model behavior,” the company says. he said.
The announced price cut means that, for example, GPT-4 Turbo is now one penny per input token, versus three cents previously, and three cents per output token, versus six cents previously – which the company calls “three times cheaper.” Tells. ” and “twice as cheap,” respectively.
Source: www.zdnet.com
