Proposal to “Re-run” The Presidential Debate

We Can Remove Biden's Cold and Lessen His Stuttering

Abstract:

A proposal to use generative AI voice encoding to change the sound quality of Joe Biden’s voice from the debate so as to allow the public to see more clearly the effects of his cold and severe stutter during the debate.

Proposal:

Many outspoken members of the media and a few in politics have said that Joe Biden’s debate performance in June demonstrably indicates he is cognitively impaired by age. The Biden Campaign and the president himself have been steadfast in their objection to this characterization. They say that the debate performance was affected by a severe cold that degraded the president’s voice, making it hoarse and weak. Objectively, this appears to be true. In addition, the president, who has a well-known stutter, appeared to have been particularly afflicted by it that night. Still, the critics contend, millions of viewers cannot be told that they didn’t see what they saw or that they can unsee it.

There, they may be wrong. It may be possible for them to unsee—or unhear, more accurately—what they think they experienced. Using generative AI voice encoding technology, we intend to test whether Joe Biden speaking his own words from the debate (via official transcript), in his usual voice, at his typical cadence, and with his usual stutter—all gleaned from recent campaign audio used to train the AI—will give some indication as to whether it is what the president said (or didn’t say) or merely how he said it as to why so many have asserted he’s in mental decline. Carefully dubbing this audio into a watermarked copy of the original video will allow us to arrive at a version of the debate where Joe Biden sounds like typical Joe Biden, rather than Joe Biden with a cold and severe stutter. What difference will it make? We can’t know in advance—it’s an experiment.

Concerns:

As is right, most reputable AI companies have strict polices against impersonation of a third party. Moreover, there is great sensitivity and caution necessary when it comes to anything election related. At first blush, this proposal might seem to be radioactive in these regards. However, it should not be. What is proposed is to use the speaker’s own voice to enunciate fairly his own words. Some fifty million viewers already saw the original debate. Many more have seen the worst clips. The value of a watermarked generative version of the debate is to openly compare it to the original, so there’s no fear of subterfuge here. A direct and open comparison to the authentic debate performance is the intent, as opposed to fooling people.

Conclusions:

The American public deserves to make up their own minds whether what they saw was cognitive decline or a technically poor debate performance by someone stricken with a very bad cold, a raspy voice, and a flaring stutter. Our proposal would be a compelling and unique application of the available technology, applied in an ethical way, to fundamentally benefit society at large, and helping to answer a pressing question with significant stakes. These are questions that could never have been answered before. Given this unique moment in time, where for the first time technology may allow it, we feel dutybound to test it.