We are in the era of “sprinkle some AI on everything.”

For this, we can thank two things. First are video games and their graphics processing units (GPU). With architectures remarkably like those of a neural net, GPUs have made computing power possible for Generative AI. The second thing is all the “killer apps” that have hit the market — Open AI’s ChatGPI and DALL·E, Midjourney, and on.

These two forces started picking up steam in 2017. Five years later, here we are.

But, through all this wonder and hype, we still have a long way to go from Generative AI to AGI.

What’s the difference between the two?

Generative AI is focused on performing specific tasks — it is all about productivity improvements.

AGI brings human-level cognitive abilities — it can understand or learn any intellectual job that a human being can.

What’s missing? What has to be added for machines to reach AGI?

1. Long-term Memory and Continual Learning
2. Personalization
3. Planning
4. Transparency
5. Conceptual Leaps

It may not take machines (and the people making them) too long until we reach AGI. Microsoft Research has published a report saying GPT‑4 could reasonably be considered a kind of early-stage proto-AGI. While Google DeepMind’s CEO says AGI may be achievable in the next few years.

Until then, we all need to keep developing our prompt engineering skills. Plus, when you get good enough at it, you can sell your work in brand new Prompt Marketplaces.

The race is accelerating.