Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Keep your AI claims in check

ftc.gov

Post reply on HN
Print view View original

281–290 of 308 posts

Re: Keep your AI claims in check

#281

Post by londons_explore »

roenxi wrote: post #275
flatline wrote: post #36

“It’s an ambiguous term with many possible definitions.” “Does the product actually use AI at all? If you think you can get away with baseless claims…” Last I checked basic optimization techniques like simulated annealing and gradient descent - as well as a host of basic statistical tools - are standard parts of an introductory AI textbook. I’ve been on the receiving end of government agency enforcement (SEC) and it…

There is also the high level question of why exactly the government needs to police this. If it turns out that some Stable Diffusion frontend was actually sending the prompts to a team of Indians who happen to draw really quickly; that is no reason to get the enforcers involved. If examined closely, the finger wagging in this post is remarkably petty. This guy was likely part of the angry crowd who didn't like Steve…

It doesn't really matter what the standard is... What matters is that there aren't some companies who push the limits far harder than others. If there are, then those companies who push the limits of what is allowed harder will be at an advantage, to the detriment of the public and the american economy as a whole.

Re: Keep your AI claims in check

#282

Post by roenxi »

londons_explore wrote: post #281
roenxi wrote: post #275

Earlier quoted context omitted.

There is also the high level question of why exactly the government needs to police this. If it turns out that some Stable Diffusion frontend was actually sending the prompts to a team of Indians who happen to draw really quickly; that is no reason to get the enforcers involved. If examined closely, the finger wagging in this post is remarkably petty. This guy was likely part of the angry crowd who didn't like Steve…

It doesn't really matter what the standard is... What matters is that there aren't some companies who push the limits far harder than others. If there are, then those companies who push the limits of what is allowed harder will be at an advantage, to the detriment of the public and the american economy as a whole.

> ...those companies who push the limits of what is allowed harder will be at an advantage, to the detriment of the public and the american economy as a whole...

Be careful with comments like that. I would remind you that [y]our performance claims would be deceptive if they lack scientific support or if they apply only to certain types of users or under certain conditions. You need adequate proof for that kind of comparative claim, too, and if such proof is impossible to get, then don’t make the claim.

If you think you can get away with baseless claims that [companies using these tactics are going to be at an advantage over companies that just make good/cheap/effective products], think again.

Re: Keep your AI claims in check

#283

Post by danbruc »

bryanrasmussen wrote: post #280
duskwuff wrote: post #182

Earlier quoted context omitted.

On the other end of the spectrum, calling a feature of a product "AI" seems to imply some minimal level of complexity. If, for example, a company marketed a toaster that "uses AI to toast your bread perfectly", I would expect that language to indicate something more sophisticated than an ordinary mechanical thermostat.

I would expect it to never burn that toast.

That would require the AI to perform at a superhuman level.

Re: Keep your AI claims in check

#285

Post by yccs27 »

10000truths wrote: post #272
klyrs wrote: post #190

Earlier quoted context omitted.

Principle problem being that expert systems required meticulous inputs from domain experts, codified by skilled engineers. People don't have time or startup capital for actual expertise...

And AI requires the same thing, we just call them data scientists and ML engineers. Using linear-ish algebra instead of decision trees doesn't change the the fact that you need time and capital to hire experts.

The big difference is that data scientists only work on the model architecture and data sources, whereas expert systems need people who have expertise in the subject matter itself. One of the biggest changes from 'old AI' to modern ML is that we no longer try to use human domain knowledge as much, instead getting the model itself to see the same pattern from data.

Re: Keep your AI claims in check

#286

Post by glenneroo »

palata wrote: post #164
dylan604 wrote: post #115

Earlier quoted context omitted.

is that a concern anymore? he's so leveraged with the Twitter purchase, i doubt he could borrow enough money to buy a Tesla

That's probably a relief for many tech employees who won't get fired for no particular reason, then :D

You must have missed the many announcements of multiple thousands of layoffs happening at many big tech companies lately ;)

Re: Keep your AI claims in check

#287

Post by y2kenny »

yccs27 wrote: post #285
10000truths wrote: post #272

Earlier quoted context omitted.

And AI requires the same thing, we just call them data scientists and ML engineers. Using linear-ish algebra instead of decision trees doesn't change the the fact that you need time and capital to hire experts.

The big difference is that data scientists only work on the model architecture and data sources, whereas expert systems need people who have expertise in the subject matter itself . One of the biggest changes from 'old AI' to modern ML is that we no longer try to use human domain knowledge as much, instead getting the model itself to see the same pattern from data.

Isn't labeling still an important part of modern AI?

Re: Keep your AI claims in check

#288

Post by Riverheart »

danielmarkbruce wrote: post #160
adamwk wrote: post #134

Earlier quoted context omitted.

In this case they aren't mutually exclusive. A government agency reminding the public it has legal authority to regulate is always threatening.

It's the casual part that is inappropriate. The government should not casually threaten the citizens.

I find your post deeply ironic because I have found a number of AI proponents to be the source of casual threats to their co-citizens ala "We're going to use AI to automate you out of a job. Better learn to adapt by paying us to use our tools yourself. If you can't hack it, sucks to suck, lol".

One company was encouraging people with a contest to take some guys work to train models and see who could best replicate it. They sent an email to the guy to let him know and say how unfortunate it was he was their target but offered to work with him to train an "official" model.

Casually threatening is how I'd describe those interactions, not this FTC post. Not suggesting that you approve either by the way.

Re: Keep your AI claims in check

#289

Post by Riverheart »

wg0 wrote: post #208
Wowfunhappy wrote: post #119

Earlier quoted context omitted.

I'm not sure I follow, I thought the article was quite articulate. Should a government agency not be proactive and articulate?

That's what I am saying exactly that it is very articulate and pro consumer. And that US is where it is because of its state apparatus. There are surely some failures here and there but it functions nevertheless. How do you read it differently?

I think the "Oh man..." and "No wonder" bits read sarcastically at first glance.

Re: Keep your AI claims in check

#290

Post by finfrastrcuture »

jsemrau wrote: post #68

We are working on an AI product in a highly regulated industry (Investing). Recently we have been experimenting using the GPT3 API as a Junior Equity Analyst. On an eyeball check the results of the technology are impressive. The problem is that there is no way to validate the feedback on scale. I.e., we can't receive statistics about the feedback from the API. In contrast, for our own Entity Recognition models we can…

A blogger (i.e. experienced ML professional in fintech) published an excellent write up of AI in fintech in December - his basic take was that there's a lot of room for the tech to grow before it becomes truly ubiquitous, because answers in finance must be correct 100% of the time. Worth a read!

https://chaosengineering.substack.com/p/artificial-intellige...

Post reply on HN