It is important to be able to verify generated media as being generated

4 Arguments

0 Citations

2 Consequences

1 Mention

Arguments

Argument #0379285c 0 0 1

If it is true that...

The rise of generative models will produce serious problems 0 0 1

Then it must be true that...

It is important to be able to verify generated media as being generated 1 0 1

Argument #18d0a8ac 0 0 1

If it is true that...

We are increasingly seeing multimodal synthetic media 0 0 1

Then it must be true that...

It is important to be able to verify generated media as being generated 1 0 1

Argument #a6cfbf73 0 0 1

If it is true that...

Synthetic media verisimilitude ‘pollutes’ commons 0 0 1

Then it must be true that...

It is important to be able to verify generated media as being generated 1 0 1

Argument #fddbb2d7 0 0 1

If it is true that...

We cannot hope to stop fake synthetic media abuse in general 0 0 1

Then it must be true that...

It is important to be able to verify generated media as being generated 1 0 1

Opposing Arguments

No opposing arguments found

Citations

No citations found

Consequences

Argument #541c3aed 0 0 0

If it is true that...

It is important to be able to verify generated media as being generated 1 0 1

Then it must be true that...

Generated media should be watermarked with an error-ensuring code 1 0 1

Argument #8eb5feb3 0 0 1

If it is true that...

Watermarking generated media with an error-ensuring code is an effective way to ensure its ability to be verified as generated 0 0 1

and

It is important to be able to verify generated media as being generated 1 0 1

Then it must be true that...

Generated media should be watermarked with an error-ensuring code 1 0 1

Mentions

Gwern/Number Search Engine

Maintained by

alexandershortt

0•

Updated

Related Propositions

With generative models, it will become entirely possible to close the loops on fabricated claims and their citations 0 0 1 Once the loops on fabricated claims and citations are closed, is it no longer enough to simply verify a watermark indicating that some text passed through OpenAI’s servers at some point 0 0 1 In trying to watermark synthetic media, we don't need to worry about preserving its truth or semantics 0 0 1 In trying to create a watermark for synthetic media, we need a system which ensures that there is always some tell-tale sign of it being fake 0 0 1 Users need the ability to know whether media is fake 0 0 1 Watermarking generated media with some tell-tale sign of it being fake allows users to manually check the legitimacy of the media 0 0 1 Watermarking generated media with an error-ensuring code is an effective way to ensure its ability to be verified as generated 0 0 1 Watermarking generated media with an error-ensuring code would fail to preserve the its truth or semantics 0 0 1 Generated media should be watermarked with an error-ensuring code 1 0 1 It is easier to create or detect locally consistent but globally inconsistent errors in synthetic media than it is to fix all the errors 0 0 1