According to OpenAI, generalized AI, in which computers think better than or at the same level as humans, is no longer a theory, and self-improving AI models are taking the world further.
OpenAI, a technology company, has just released GPT-5.5, a new artificial intelligence (AI) model touted as the best to date in handling research tasks such as creating improved versions of itself, amidst a rapid pace of new product launches by rival AI companies.
GPT-5.5 was considered a "new class of intelligence" and was launched just a few months after its predecessor was introduced.
OpenAI has developed the ChatGPT chatbot, a model particularly strong in proactive programming and computer usage, where independent digital assistants perform tasks as instructed.
According to OpenAI, generalized AI, in which computers think better than or at the same level as humans, is no longer a theory, and AI models that study how to improve themselves are moving the world further in that direction.
Executives describe GPT-5.5 as one of the most clear steps to date toward models that can accelerate AI research itself.
Greg Brockman, co-founder and chairman of OpenAI, said what's truly remarkable is that GPT-5.5 can do more with less instruction. This model is laying the groundwork for how humans will perform tasks on computers in the future.
Meanwhile, research director Mark Chen said that, in the short term, OpenAI focuses on having humans play the role of "coordinator" while AI models do the "heavy work."
OpenAI asserts that, with GPT-5.5, they have built the strongest safeguards to mitigate abuse, especially against biological and network capabilities.