Is AI + Databases the future of computing?
If you think about how knowledge works, you can potentially divide it into 3 areas:
1. Fact based
2. A mix of fact based + Imaginary
3. Imaginary
Here imaginary is used to represent a wide range of things — from completely hypothetical to fact based that do not apply contextually.
Now, what’s very interesting to me is that most humans can switch between all three modes effortlessly.
When you are asked what’s the date on the first day of the year, I’m sure most of you all can quickly answer January 1. When you are asked to tell a really creative story, you can spin one up quickly. And when you’re applying for leave, you’re probably taking some creative liberties while describing your situation (fact based + imaginary).
Now here’s what I think has changed since the beginning of the LLM revolution.
We always knew computers were very good at fact based retrieval.
Heck, they’re even better than us and remember everything! So, in a way, a database has traditionally been a source of fact based knowledge.
What’s different with LLMS is that LLMs (or AI as it is popularly known) bring the other dimensions into play while being somewhat weak on the first point.
Multiple ChatGPT demos showcase the capability to hallucinate and therefore produce extremely creative stories. Additionally, the outputs can also be guided via prompts to make them more realistic.
But often, where they have been known to have issues is getting fact based answers right.
However, interestingly, this is something that computers are already generally good at (databases and other sources of data).
Considering the above points, the future is likely to be a hybrid of traditional database technology and AI/LLM based new age technologies. And this will solve for many problems today and usher in a new era of computational progress!
