00:11 — This episode is brought to you by the Cloud Wars Expo. This in-person event will be held June 28th to 30th at the Moscone Center in San Francisco, California.
00:36 — Massive data sources are needed for AI to learn from. However, this creates massive AI Models.
00:56 — For example, in 2020, OpenAI launched GPT-3, which is a large languages model with 175 billion parameters. This AI model writes short stories in human-like writing styles when prompted with a single sentence. Although this is a major step forward with AI and human language, it still can’t create anything new.
01:59 — In December, DeepMind released an AI model with 280 billion parameters. Megatron and LG released an AI model with 530 billion parameters. Then, Google released its Switch Transformer and GLaM models with 1.2 trillion parameters.
02:31 — Despite these advances, these AI models are still prone to bias and toxic outputs.
02:54 — As humans are using AI more as part of our daily lives, there seems to be a gap with purposeful usage and adoption within businesses.
03:37 — We have to create a balance between what AI is outputting and human ingenuity, experience, and creativity.
Looking for real-world insights into AI? Subscribe to the Enterprise AI Impact channel: