Opinion: Is open source AI the only trustworthy long-term way to develop AI?

While OpenAI does not expect to be profitable before 2029, the startup’s valuation keeps climbing in funding rounds baffling some financial analysts – Copyright AFP Kirill KUDRYAVTSEV

Unlike just about every new class of technology in history, AI has a unique ability to be universally despised. The ridiculous investment racket, combined with the actual performance and delivery of the product, makes it easy.

In coding terms, AI often looks more like slapstick comedy. The AI Keystone Codes go well with the Keystone Cops.

Inept, unreliable, farcical, and hallucinating software is nobody’s idea of a credible standard of performance. Certainly not business performance. One lousy prompt or a few random otherwise unemployable idiots can produce a MechaHitler.

This is your idea of a few trillion dollars’ worth of good investment? Ownership of this extremely expensive bad case of poultry diarrhea could be legal suicide in one process.

It’s called AI slop simply because it is sloppy.  

Imagine this dialogue with an AI agent:

“What’s our current electricity load on the grid?” you ask hopefully.

“Oh, I dunno. How about a meaningless rabid political rant instead?” it replies with a simpering and generally inexcusable New York accent.

Great stuff.

No, it damn well isn’t.

Criticisms of turgid AI videos, endless repetitions of the same information, and the content curation of a dunghill are all perfectly valid and correct.

There’s no actual value in any of this garbage.

Now, let’s get a little less superficial.

This generation of ultra-smug AI isn’t and can’t be the whole story, thank god. Even allowing for the Rectal Rhapsodies of AI spruikers, it’s an interim stage before high-functioning AI can evolve.

This is where the heavy-duty professional criticism necessarily kicks in. it’s interesting to note that in the IT sector, where the real high-tech guys reproduce by fission, none of the AI BS flies at all.  

These are the guys who evangelized the internet and every single electronic component since the Commodore 64 like proud parents. They know how derivative current AI is, glued on to much older software like writing and music software that is often decades older.

For the first time ever, the Ultra-Geeks and consumers are on the same page, if for some similar and some different reasons.

They agree that:

Big Tech is all about money and not about performance.

Corporate agendas are driving development for purely financial reasons.

Added ornaments like rumoured new hardware requirements for Windows 12 and everything else, and similar bric-a-brac are more obstacles than assets.  

All of this is being done at a great distance from consumer needs, and the consumers don’t like it.

Why not put this externally sorta-maybe required but not consumer-essential junk on the Cloud, where it belongs?

This unsightly mess brings us elegantly if verbosely to the issue of open source AI development.

Open source development is a comparatively complex idea, but it’s essentially free. Open source has generated some of the most useful software ever developed.

The big issue for AI is that open source is also transparent. It can’t really be a corporate toy. There are overlaps, sure, but that’s inevitable. The intellectual property game, however, is very different.

Real development happens in people’s heads.

Some poor soul, floundering through cumbersome protocols and obscure IT idiocies will notice that there are workarounds. Anyone who’s ever done coding will tell you that workarounds are the main reasons that anything works at all. That’s pretty right.

Now, a bit of logic:

If corporate agendas distort efficient development, or more likely sidetrack it, open source is the way around those agendas.

Pure research, the only real gold standard for real tech breakthroughs, is unrestricted in the open source environment.

Talent isn’t and can’t be confined to a stingy 5-minute input in some useless meeting.

Open source is peer-reviewed and highly visible by definition. Its trustworthiness is based on actual performance, not some damn PR exercise.

Open source will finally free AI development. It’s that simple.

______________________________________________________________

Disclaimer
The opinions expressed in this Op-Ed are those of the author. They do not purport to reflect the opinions or views of the Digital Journal or its members.