OpenAI DevDay Preview: What to Look For on Monday
OpenAI is hosting its first developer conference on Monday, Nov. 6, with a keynote at 1 p.m EST, 10 a.m. PST, and there’s a lot of anticipation over what CEO Sam Altman and his team will present during the event. The OpenAI DevDay announcement only hinted at “new tools” and plans for breakout sessions with the company’s technical staff, but there are some signs that some big features and upgrades will be unveiled, including better tools for designing generative AI applications and possibly autonomous AI agents.
OpenAI’s Developer Boost
openai devday is in a week–11/6 at 10 am
we have some great new stuff for you!
will be livestreamed on https://t.co/VaGG9g2o5x
— Sam Altman (@sama) October 30, 2023
The big question is whether the Artificial General Intelligence (AGI) Altman and OpenAI have brought up many times will make an appearance. The current rumor is that an autonomous agent supposedly code-named Jarvis will serve as a demonstration. The company recently rewrote some of the text in its ‘core values’ statement to place AGI as the paramount goal for the company. It would also clarify the thinking behind OpenAI’s purchase of digital studio Global Illumination as OpenAI could train its AGI-fueled agents through a gamified version of Global Illumination’s digital world. AI agents would also benefit from the recent move to bundle ChatGPT’s tools into a single setting.
Another centerpiece is likely to be the addition of memory storage capabilities to the developer platform, using what is referred to as a ‘stateful API’ that can refer to a user’s conversational history with a conversational AI. This will allow large language models like GPT-4 to remember the context from previous conversations. Developers employing this feature might end up slashing their computing costs by a huge amount. That’s crucial when usage fees are a common bottleneck for a business adopting generative AI. The conference may also include APIs of some of the new features introduced in the last few months to ChatGPT, including visual processing. As with the generative AI chatbot, developers will be able to embed ‘eyes’ that can understand photos and other images well enough to describe them and analyze their content.
Voicebot will be hosting a live discussion after the keynote. Check back for more details soon.