sriniwasx

Specialized chips like what cereberus is building combined with small models can give great result in the long run, but at the pace AI is growing it's not very practical yet.

Looking at the trends, local AI is the future, smaller models are getting faster and high demands always leads to oversupply if you've got resources, just look at all the AI data-centers, If small models go mainstream GPU's and data-centers will become dirt-cheap.

As cool as it sound, do we actually need it? also what problem are we solving?

Performance is probably the only angle that has some potential but only short requests, the db and other external IO/services is a bigger bottleneck.

With the models we got at out disposal, you can easily deploy a kickass internal tool, it's fine even for mvp's unless you're not handling sensitive user data, shooting yourself in the foot is easier than ever, and talking about code smell for most application (especially frontend) does it even matter?

The code smell in my repos are at an all time high and I'm a senior dev, can't image how worse vibe coders have it.