The entire market seems to care about data rather than privacy.

From your smartphones to your cars to your watches to everything else, it seems like the economy is about getting as much information about you to advertisers/corporations/governments. And data collection looks to be accelerating if anything.

It's like the system is giving us the rope and we are mindlessly putting it around our necks.

The orwellian nightmare was that big brother was going to forcibly install telescreens and listening devices all over your home to deprive you of an ounce of privacy. He never imagined it'd be us installing the telescreens and microphones all over our homes to deprive ourselves of privacy.

Hey man, when I was working in that field, we had really strict controls on who could access the data. I had to sign A PIECE OF PAPER agreeing not to misuse customers' personal data before I could get at it. Plus I'm pretty sure the company did a cursory background check before they hired me. And there's a form that government intelligence employees have to click through sometimes before they get the data.

Privacy secured.

Hey, good for you! But how am I to verify that other companies take the same level of care?

Also, just because engineers have to jump through hoops to be able to run SELECT on a database, doesn't mean marketroids and sales don't get free rein on using it (or passing it to other parties).

I think your sarcasm detector may be broken.

At what point did the market care for that?

There's a niche market for such products, in which a high premium would be paid for a product that successfully meets the needs of its purchasers. Unfortunately, successfully is defined by the high floor of convenience provided by multibillion dollar companies, so I understand that it's tough to compete with that.

I've participated in adjacent niche markets where consumers 1) most likely work in tech 2) have needs that 99.9% of people don't care about 3) are willing to pay a premium 4) have the means to pay that premium.

It's too bad that there are so few businesses willing to tap into this market. Part of me believes this is a marketing issue, though. I've yet to see a campaign that really nails our core concerns and does so elegantly.

You create it for the part of the corporate market that wants good on prem services. On-prem that ends up also being something for prosumers is basically the missing thing we are looking for, kind of like ubiquity wifi.

How do you propose growing representative training examples if they can’t collect the data, at least in part?

Would depend on my budget (because it would certainly cost more than hovering up all training data for free).

Perhaps: Pay/reward for data sharing, decentralized training, differential privacy, local training and submitting the weights, local fine-tuning of pre-trained model, marketplace of third-party (open source) models, ...

... or just have the user repeat some phrases for 5-15 minutes during initial setup? Like you could do with good ol' Microsoft SpeechAPI from the first decade of this century?

The obsession with cloud-based machine learning is creating half of these problems.