ChatGPT: Next version will be 10,000 times more efficient
ChatGPT works wonders, to the point where OpenAI’s artificial intelligence has been in the spotlight in recent weeks. Recently, conversational AI was surprised to “help” many students at a university in France to write a written work.
We also noticed that social networks echo ChatGPT’s more or less sophisticated responses to the most diverse (and sometimes the most absurd) questions. This technology is based on GPT-3, an advanced language model based on machine learning. This model is destined to progress with updates, an often spectacular progression from one version of the software to another.
GPT-4 is already in development and OpenAI is expected to release it later this year. The difference between GPT-3 and GPT-4 promises to be truly gigantic: To the point that GPT-4 is capable of managing 100 billion parameters, against “only” 175 billion parameters for GPT-3, that is, GPT- 4 would be “approximately” 10,000 times more powerful than GPT-3, which should allow it to offer new types of services, considerably increasing its reliability and its resistance to erroneous data.
Programmer Simon Høiberg reveals that he even believes that the release of GPT-4 “will make ChatGPT look like a toy”.
Newer Articles
- Twitter: Several third-party apps mysteriously stopped working
- Twitter blocks Third-party Applications
- WhatsApp latest feature will let users Block annoying Contacts
