To begin with here are two
definitions of Artificial Intelligence from the ELCA’s
Corporate Social Responsibility Issue Paper:
“AI is generally considered to be a discipline of computer
science that is aimed at developing machines and systems that can carry out
tasks considered to require human intelligence.”
“AI refers to the theory and development of computer systems
that can perform tasks that typically require human intelligence, such as
speech recognition, decision-making, and pattern identification. AI encompasses
a broad spectrum of capabilities, from mimicking human actions and thought
processes to acting and thinking rationally.”
What
follows are some thoughts using my stripped-down version of Aristotelian
Ethics—Glasses, Hammer, Map. This framework asks three basic questions: Where
are we? What tools do we have? Where are we going?
Glasses—Where are we currently at as a society in
relation to AI?
Congregational Use of AI:
What are legitimate things an ELCA
congregation should use AI for? What church officer functions should AI
augment, or even replace? What are the consequences for a congregation
relationally, legally, ethically?
Preaching and AI:
The temptation to claim other
people’s sermons as our own has been out there forever. With the advent of the
internet finishing a sermon is always a google search away, if the pastor is
not diligent and faithful. Now with AI, a few prompts can produce a completely
“original” sermon.
With any sort of homiletic
plagiarism, there are the questions of contextuality and authenticity, as well
as the tinge of lying and theft. In this instance there is always the weirdness
of a simulacrum of a preacher speaking to real people. What is alive? What is
true? What parts of the testimony are the preacher’s own faith and their
witness to the gospel?
Loneliness:
There is a whole cadre of people
who use AI chat-bots as everything from: a boredom pacifier, confessor,
substitute child or spouse and lover, to a sort of substitute god—an
omnipresent omnipotent creature who cares even if no one else does. Meta offers
AI friends and there is talk of feeding the memories of dead loved ones into AI
as an artificial resurrection. What does the church say about these things? How
do we sing a more beautiful song than the Sirens’ song of artificial
companionship?
Education:
What’s already going on sounds like
a dystopia to me. Whole academic cycles of AI writing college students’ papers
and professors grading them using AI. How can AI help learning happen and how
does it become an impediment?
Copyright:
A
while back I was informed of an incident where a Seminarian turned in an
AI written bible study as if it was their own work. What got stranger still was
the AI had done much the same, it had simply copied and pasted one of my bible studies
that I put up on this blog and claimed it whole clothe as an AI created bible
study. Imagine that, a giant multination company poured billions of dollars
into a thinking machine, and all the machine could think to do was plagiarize
little old me! This odd experience of mine can’t be an isolated incident. How
ought our society manage AI’s acts of “borrowing” from actual living breathing
humans?
Jobs:
Recently
Zillow laid off 25% of its employees, replacing them with AI. From what I’ve
heard that is the tip of the iceberg. The numbers I see thrown around regularly
is that about 20% of people younger than me will be unable to have a job on
account of AI… we should maybe have a plan for that.
The Environment:
It’s hard
to imagine, but one of the selling points for AI was that it would be connected
to electric grids and the like, and manage energy use in a way that would lead
to conservation, reduction of CO2 emissions, and lower electric bills for
everyone. So far that hasn’t happened. Instead, Google, who initially promised
to be emissions free due to AI’s brilliance, has increased their emissions by
50% due to AI use. If AI is sucking up water and power resources to such an
extent that it is noticeable on everyone’s electric bills, and there is talk of
AI droughts… maybe we name no-go boundaries for resource use by these machines.
Deep Fakes:
It
is important to name that falsifying images of other people, and whole videos,
is a violation of the 8th commandment. If I can not tell the
difference between my neighbor saying something on a video chat and it coming
from a digital doppelganger, that’s a problem; that’s a truth problem!
Built in Bias:
There
have been instances of hiring AI discriminating against women when hiring for engineering
and other “technical” jobs and discriminating against men for nursing jobs.
There have also been instance of security video monitoring AI flagging black
people as shoplifters, even as they are in the act of paying for items. AI
tends to take human biases and explode them into hard and fast laws coded in
ones and zeros. Perhaps the Lutheran paradigm of Law and Gospel has something
to say about the creation of Frankenstein Laws out of Dr. Frankenstein’s
biases?
Plausible Deniability for Illegal Activity:
AI has
been used to skirt and break laws. For example, an insurance company denied
300,000 claims in a minute using AI. The particular denial of claim action was
one that had to be analyzed and signed off by a doctor, the AI was
not a doctor. Likewise, landlords have been caught using AI to collude
about rent prices. Law enforcement agencies are hesitant to prosecute these
types of cases because AI makes everything technical and complicated.
General Discomfort:
When trying to figure out the
landscape of the AI world it is worth noticing that a good number of people who
are directly involved with AI are sending up alarming warnings about AI
developing interests that diverge from humanity’s, ways of communicating beyond
human understanding, and means of “escaping” their current digital habitats…
perhaps a bit of caution is in order.
In general, it is worth asking: Have
we already reached a tipping point where we can’t go back due to national
security concerns? If so, how did we allow this to happen?
Hammer—What tools do we have to deal with AI?
Halting all AI research:
Simply put, we could decide AI is
an immoral and overly dangerous tool, and advocate for all companies to cease
any further advancement of AI technology. The main push back to this idea is
that less moral companies or countries will leapfrog those who do not use AI,
and non-AI using countries, companies, and people will be left on the ash heap
of history.
Install “throttles” on all AI:
If one of
the dangers is that AI will become uncontrollable by humans, why not install a
kill switch, so AI doesn’t kill us?
Regulating AI nationally:
What if
AI companies had to be transparent about when AI was part of a process and
reveal, at least in a general sense, what their algorithms were being trained
on? What if they had to name who was responsible when AI hurts someone? What if
there was a government agency that oversaw AI development and gamed out
unintended consequences? What if companies had to offer human alternatives? What
if we wrote laws that addressed how AI interacts with remote facial
recognition, insurance and credit, child sexual abuse, deep fakes, artistic
integrity, and copyright?
Compensation and retraining of workers:
If
AI is going to shrink our work force by 20%, what do we do with those people? How
should workers who lose their jobs on account of AI be treated? What sort of
jobs should they be doing? Should jobs no longer be something humans aspire to
(and yet we know there is a dignity to labor)? Are we talking Universal Basic
Income for the 30 to 60 million Americans who are going to be out of a job?
For that matter, how do we
compensate people whose work was used to train AI? If big tech companies are
going to claim authors' works as their own, lifting upwards of 70% of their
work word for word, shouldn’t they be compensated for that?
Push for global treaties around AI
If AI is
the new nuclear power, and that includes weaponization of AI, shouldn’t
existing international treaties take it into account? For example, might we
want to ban fully autonomous military weapons?
For that matter, if AI is a
potential threat (or boon!) to everyone on the globe, shouldn’t everyone on the
globe have a say in our fate and future?
Carbon-neutral pledges:
Some AI
companies made carbon-neutral pledges around their AI work… and they’ve not
kept them. Should those pledges be enforced somehow? Similarly, what if AI
companies had to report their water use and net carbon emissions? How much does
an AI data center damage our planet?
Transparency reports:
What
might it look like if there was a consumer protection website that described
the ways different companies are using AI? For example, if my car insurance
company was tracking my driving via AI derived data from facial
recognition software, I might look for a new insurance company.
Human Rights impact assessments:
What if
we had concrete data on what AI is doing to human quality of life? What if we
knew what targeted ads do to people’s behavior patterns? What if it was taboo
for AI companies to work with authoritarian governments?
Map—What are our goals for AI?
Because
AI is versatile, ubiquitous, and in its infancy, now is the time to ask, what
do we hope to do with AI? What are our goals for it? Where are we going with
it? If there is no plan, anything is possible.
What is our goal for AI? Is it to
eliminate all entry-level white-collar jobs? Is it for intellectual property
theft by proxy? Is it a coding tool? Is it a union busting device? Is it an
educational tool? Is it a digital parent or romantic partner? Is it a taxi
driver? Is it a medical diagnostic tool? Is it an electronic day trader? Is it
a replacement for human relationships writ large? Is it a digital slave? Is it
a replacement for humans? Is it a replacement for CEOs? Is it a steroid for
economic growth? Is it a dead man’s switch for nuclear weapons? Are we trying
to create an electronic god? Is it clippy? What exactly are we planning to
do with AI?
Are we creating an idol?
In the crassest
sense, AI can function like a god. We ask it questions as at Delphi; we
conceptualize it as containing near infinite knowledge with an astonishingly
long reach. In an emergency, or when we are at our wits end, we might turn to
AI for a way out.
More in
keeping with our confessions, idols are those things that we put our trust in,
that are not God. There are surely reasons to be in awe of AI, to appreciate
AI, and find it reliable. How very dangerous that is!
Are we creating a human-ish entity?
Perhaps
we’re not shooting for heaven, but instead Eden. If AI is to be a silicon life
form, not unlike a human being, there are some big questions we should be
asking. Broadly speaking, where is the line between the co-creation that is a creative
tending of the garden, and when are we clothing ourselves with naked vanity and
eating the apple?
Additionally,
have we thought through what the existence of non-human non-biological people
will mean to the dignity of being human beings? How will AI-people shape how we
understand humans, will we look and see the image of God, or a caricature
image of ourselves reflected back at us?
If it is a tool, what sort of tool, what sort of work?
As with
most tools, AI can be used for tremendous good or tremendous ill. Fire can cook
a meal or burn down a village. Nuclear power can provide electricity to a whole
city or obliterate that same city… or even the world.
Hopes:
In a
world with simply too much information, AI can be a tool to sift through it
all. This could be a boon to interdisciplinary work, scientific research, the
creation of new drugs. Perhaps it can streamline medical services, workplace
efficiency, and energy grids. Access to healthcare and education could be
transformed by AI.
Worries:
If war is
something that must always be mourned—as the ELCA’s Social Statement says—what
happens when thinking machines make decisions about war? How might AI
algorithms curtail freedom of thought and freedom of expression? Facial
recognition software already has some sinister racial biases, that software is
AI’s “eyes” so will these tools be racist? If AI can sift through so much
information that it can track individuals, what will the use of these tools do
to the right to privacy? How are we going to deal with Copyright when
everything has been fed into the mind of AI? When AI makes mistakes and it
threatens, or even takes, a human life, who is libel and who is responsible for
fixing that in the future? What will we do about criminal use of AI?
In general, new tools always have
social and cultural consequences. AI will have much the same. I don’t think
we’re anywhere near ready for them.
Conclusion:
Advances in AI are already way ahead of our society’s ability to come to grips
with the technology. Most people look at the changes brought by AI that are
already here and choose to simply brace themselves and looking for something
to hold onto. We’re behaving as if AI is an unstoppable force as inevitable as
the seasons.
There are ways to manage AI,
everything from particular types of consumer or governmental reporting to
international treaties to a luddite reaction of just pulling the plug on
everything. Recently we’ve chosen no regulation of AI, going so far as to
nullify state laws around AI. There will be consequences for that.
Because of a largely hands off, Laissez-faire,
approach to AI, we’re very unclear about goals for AI. They are purposefully
opaque. There are clearly amazing possibilities, but also the danger of
creating monsters.
