[ExI] Tools, Intent and Culpability (was Re: 23andme again)

Kelly Anderson kellycoinguy at gmail.com
Thu Jun 27 17:05:47 UTC 2013

On Wed, Jun 26, 2013 at 7:03 PM, spike <spike at rainier66.com> wrote:

> This episode has pointed out to me an ethical blind spot have had for years
> and never really pondered, one which is related to AI and singularity
> research.  I will participate in development of software which performs
> duties I would not do myself.  I did not and would not tell the
> information,
> but I pointed out how to use existing software to get the same info.  So I
> will not be the bad guy myself, but I will explain how to use software
> which
> will do the same thing I wasn't willing to do, or even let my own computer
> code be the bad guy.

As pattern-seeking, tool-making primates, I'm sure our ancestors have
debated the ethical issues surrounding the use of tools since before the
capture of fire. There is a rather common consensus that the maker of a
tool has zero culpability for the use of said tool.


So in the above (fascinating and G rated) picture, one could ask, "What is
the culpability of the maker of the horse shoes in the murder? I believe
most rational people would say that the culpability of the blacksmith is
zero. Horse shoes are not made with the intention of being used for murder.

Similarly, the manufacturer of a knife is not held responsible for murders,
even though knives are much more suited to this purpose than horse shoes.

If you managed to commit murder with your computer, the manufacturer would
not be held responsible, even though computers can be used to do
practically anything.

Then we come to hand guns and assault weapons. Here, for some reason that I
cannot begin to fathom, the rules are different. Yes, an argument can be
made that a gun is designed to inflict harm, but so is a whip. There are
movements afoot to find gun manufacturers responsible for murders committed
with their guns. But this makes no sense.

They conflate these arguments with those about cigarette companies. The
makers of cigarettes, however, had information that their product was
harmful for its intended use, and intentionally mislead the public that it
was indeed safe.

I see no comparison. A gun is usually purchased for play, hunting or
protection. The gun does its job. If the gun backfires and injures the
user, that is the case where a law suit might be appropriate. The gun did
something it wasn't intended to do.

Now, getting back to 23andme and computer programs. The harm done by such
mechanisms is tertiary to the intended positive outcome of providing people
with fun and possibly useful medical information. The negative outcomes are
much smaller than the positive ones.

As Kevin Kelly said in "What Technology Wants", if things get 51% better
and 49% worse, that still makes things 100% better every 35 years or so. We
must accept the negative uses of tools like Google along with the positive

There are cases where I wonder if the positive outweighs the negative. For
example, in doing facial recognition on every picture and video on the
Internet. That seems like a case where the negative might well outweigh the
positive in loss of privacy and potentially very negative outcomes indeed.

So while I think some hand wringing is valid, the cases where the negative
outweigh the positive are either rare or obvious.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20130627/292452d7/attachment.html>

More information about the extropy-chat mailing list