ChatGPT 3.5 still falls far short of GPT-4, and other models have long exceeded it.
On Monday, OpenAI announced that users of the ChatGPT website in some locations may now utilize the AI assistant without registering. Previously, users were forced to register an account to utilize ChatGPT, even with the free version, which is now powered by the GPT-3.5 AI language model. However, as we previously stated, GPT-3.5 is well recognized for delivering more erroneous information than GPT-4 Turbo, which is accessible in commercial versions of ChatGPT.
Chatgpt
ChatGPT, which first debuted in November 2022, has evolved from a technical demo to a full AI assistant with a free version available at all times. As the saying goes, “You’re the product,” hence there is no price. ChatGPT enables OpenAI to gather data that will be utilized in developing future AI models; nevertheless, both free users and ChatGPT Plus subscription members have the option to opt out of AI training. (OpenAI has stated that it will never learn from the ChatGPT Team or Enterprise members).
Opening ChatGPT to everyone may provide a frictionless on-ramp for users who may use it as an alternative for Google Search, as well as potentially generate new customers by making it easy for people to use ChatGPT quickly and then giving an upsell to premium versions of the service.
“It is central to our mission to make tools like ChatGPT widely available so that people can reap the benefits of AI,” the company’s blog states. “For anyone that has been curious about AI’s potential but didn’t want to go through the steps to set up an account, start using ChatGPT today.”
Because children will be able to use ChatGPT without an account, which is against the rules of service, OpenAI claims it will implement “additional content safeguards,” such as barring more prompts and “generations in a wider range of categories.” OpenAI has not clarified what this implies, but we contacted the business for a reply.
There might be some potential drawbacks to the open approach. On X, AI researcher Simon Willison commented on the possibility of automated misuse as a method to avoid paying for OpenAI’s services: “I’m curious how their scraping prevention works. I expect the temptation to misuse this as a free 3.5 API would be rather great.”
With intense rivalry, greater GPT-3.5 access might backfire.
Willison also mentioned a common criticism of OpenAI (as voiced in this case by Wharton professor Ethan Mollick): people’s ideas about what AI models can do have so far been largely influenced by GPT-3.5, which, as we mentioned, is far less capable and far more prone to making things up than the paid version of ChatGPT.
“In every group I speak to, from business executives to scientists, including a group of very accomplished people in Silicon Valley last night, much less than 20% of the crowd has even tried a GPT-4 class model,” wrote Mollick in a tweet from early March.
With models like Google Gemini Pro 1.5 and Anthropic Claude 3 possibly outperforming OpenAI’s best proprietary model at the moment, with AI models outperforming the free version of ChatGPT, enabling users to utilize GPT-3.5 may not be the greatest way for OpenAI to present itself. Microsoft Copilot, which is driven by OpenAI models, likewise provides a seamless, no-login experience, but it only enables access to a GPT-4 model. However, Gemini now requires a sign-in, and Anthropic provides a login number by email.
For now, OpenAI claims the login-free version of ChatGPT is not yet available to everyone, but it will be shortly. “We’re rolling this out gradually, to make AI accessible to anyone curious about its capabilities.”
