I legit, cannot figure out what you question actually is.


if this is it, I absolutely disagree unless again you are only saying something like removing **** every time it is written and instead making us all write f*ck, is what you are trying to define as the substantial reduction.


If you are saying on this forum alone F*ck would have been written and was denied with ****, that is just a silly point.

The heart of this issue is people who will not obey TOS and have an agenda in their posting that they would post in denial of TOS and that human moderation is the main thing to remove it. No meaningful reduction at all would be made from anyone intent to post things that would break the rules and that is what would matter to the site owners as the ones liable.


?

The question I would ask an agreed on expert is:

"What impact could AI make on preventing harmful content being posted on facebook over the next 5 to 10 years"?

We would then agree on someone like T_D to determine if the answer amounted to 'not significant' when I would lose the bet. You would be nuts to take that bet but I dont think you would agree that question because you're talking about something else.

Ya I am not sure what the range of answers you would expect are but I know they would all be very heavily hedged.

If you broke that question up into:

- do we currently have in AI any real ability to prevent harmful content being posted on SM - That answer would be a flat 'NO'
- do we feel we are in sight of a break through currently with any of the AI work being done in this area - that answer would be 'NO'
- do you believe with all the money being poured in we could get the type of breakthru that would achieve that - that answer would be 'it certainly is within the realm of possibility', 'I am cautiously optimistic
- would you bet that within 5 or 10 years that break through would be made - Well 10 years is better than 5 but it is not something I would choose to bet on. There is a significant percent of the community who believe this breakthru may never be achieved


NOw on both sides of that you would find boosters and deniers who would take very strong positions, but the bulk of the AI scientists and the bulk of AI conferences addressing this issue would take my position above. It is the position I have seen discussed at conference panels with most of the worlds top AI theorists and scientists.

Yur just reintroducing the confusion over the word 'prevent'. Of course we have tech capable of preventing some harmful posts.

I'll stick with my 'bet'. I think the meanings are clear and I'd happily give you all the answers that T_D marked as a 'not clear'. I'd also happily except that I need 5+ out of 6 to win.

If you would really take that bet in pricniple then at least we now know what we disagree about.

btw I did't say 'over the next 5-10 years' to allow for breakthroughs. That was to allow for implementation, training and squabbles over the regulations.

No we do not and we will never agree on that. What we have is the abiilty to push people to use another form. Various equivalents of technology forcing us to write f*ck instead of ****.

That is where we differ as you say 'look' it blocked you saying **** thousands of time thus success. Whereas I say people just wrote 'f*ck' thousands of times instead which is in no meaningful way any different, if the law says you have to prevent the intent' and the 'underlying message' and not just the words.


You obviously think the meaning is clear and I don't.

To me your bet question makes no sense. It would be like me telling you I am 50/50 on intelligent life existing and you saying you want to bet me that you think some expert or mediator would disagree with me and say they do exist, and you saying you are confident you would win.

There simply is no bet there for me to take even if you are super confident they do exist and you think others would agree with you. You have not frames a question or bet in any way I could take it. You are speaking gobbledygook.

ok. The 'bet' I outlined is s so simple and directtly related to the issue of making a significant reduction in harmful posts that let's just leave it there.

As I've pointed out in that discussion with Candybar we sometimes reach a point where there is a compete disconnect that I cannot fathom

Another part of the suggested regulations

https://www.gov.uk/government/news/n...-trolls-online
Doesn't need AI to be able to make a massive difference in preventing harmful content but AI will play a part in catching those trying to fake verification.

I suspect in some years time there will a requirement to make it a very clear option or even the default setting.

At first I thought that one sounded like a good and practical thing to do but on second thought I think it will be largely meaningless.

I think 99% of current users on those platforms (outside celebrities and influencers and a small group who claim their identity) will opt to remain unverified and thus not utilize the tool of blocking the other 99%, who are not them, only to interact with the 1% who are verified.

As you say, this is an area AI might be able to help where some type of "trust' or 'Verification' score is built that can approximate someone verifying the old fashion way with enough confidence that people then would use it.

I also wonder if the major platforms could entice verification. Find some way, to maybe reward the verified, with a level of user experiences ('Gold Membership') ? Maybe get corporate partners to create gift bags of products each month to be given out randomly amongst the verified and in return the corporate partners maybe get the 'verified data', the gold mine of corporate data collection.

I don't think the SM platforms could require people verify without then creating massive losses to platforms that do not.