Ann: Appen teams with NVIDIA to enable AI solutions, page-14

Chat GPT 3 has 175 billion parameters only systems like Deepmind Chinchilla with 70 billion parameters but more extensive data trainings can outperform Chat GPT 3.
Chat GPT 4 has 1 trillion parameters. model size, data quality and the amount of data trained are critical to AI performance - parameters in language models is the ability to compute complex patterns and relationships between training data and develop new capabilities alongside text. Parameters determine the relationship between input and output through weights. The weights are learned during data training.
Everything else is just utopian wannabe magic just like unsupervised A.I being able to determine the difference between a giraffe and goat without data. A.I needs alot of data to be accurate and commercially viable everything else is nonsense.