“This book is my last battle in my lifelong mission to fight devastating ignorance…. Previously I armed myself with huge data sets, eye-opening software, an energetic lecturing style, and a Swedish bayonet for sword-swallowing. It wasn’t enough.” — Hans Rosling, Factfulness
Replika is an AI chatbot “whose sole purpose is to become your friend.” Designed by a Belarusian woman, Eugenia Kuyda, who also created Luka, a forerunner to Replika, when her friend died in a car accident. Luka was a tool, built from the deceased’s words in texts, that would allow her to communicate with her dead friend’s ghost.
Luka and Replika — aside from their profound cultural and psychosocial implications — are artifacts* built from facts. The facts and words of a deceased friend’s existence; the facts of your life that you “tell” a chatbot which has been programmed to commit acts — messages — of friendship.
Linguistically, artifact adds art to fact: “anything made or modified by human art.” (From Italian artefatto, from Latin arte “by skill”+ factum “thing made,” from facere “to make, do.”) Historically, artifacts have been artistically conceived and executed approximations: drawing, paintings, written narratives, b & w silent movies. With the advent of mirrors (see David Hockney’s Secret Knowledge) and later photography and photographic reproduction, artists and technologists have worked to effect ever greater realism in artifacts. Today with Photoshop and the advent of deepfakes, we have arrived at a curious juncture: perhaps due to mirror neurons, perhaps due to other as yet unexplored reasons, humans want their artifacts to look and seem real. As Umberto Eco wrote: “the American imagination demands the real thing and, to attain it, must fabricate the absolute fake ….”
Artifacts now have, or soon will have, the reality of facts and objects in the so-called real world. These now, or soon will, compete with the facticity of a thing. In effect, we are creating artifacts that look and act so real we cannot tell which is real and which is not. Further, the conflation of fact and artifact puts our understanding of “real” and facticity up for grabs.
We are awash in facts. We count our steps; we count our calories. Our newscasts keep count of Covid infections and deaths. We know the scores of all professional and college sports from the Montreal Canadiens to South Texas women’s rugby; we watch moment-to-moment market fluctuations, highs, lows, and averages from Tokyo to the Nasdaq.
We now have more facts about more things in more places delivered to us by more devices in more hands — than ever before. But our presentation of facts — our digital artifacts — can manipulate and distort facts in ways we do not yet fully understand. Regarding device logic — and our entraining with that logic — when we pick up and use new tools, we are seeing new ways of behaving in the world. We forget that using the alphabet as the sole tool or intermediary between ourselves and the world changed once computers entered every object on earth and cell phones and algorithms brought us Twitter and TikTok. The world is now … what brings us the world.
Artifacts are enlivening, creating another life form, a sensible Metalife. Our urgent need now is to link artifacts to facts; to ensure artifacts do not, as they are designed to, and are able to, take off on their own; become untethered from what is real, what is true. And as proof of the breadth of this phenomenon, you need only look at your own life and the information the world has about you. Any fact about you can become, or is now, an artifact. It’s all living out there somewhere. The larger issue is our blindness to the unnoticed level that artifacts represent.
We must begin to acknowledge how artifacts have captured, and thereby diverted, our attention. Let’s start with the basics.
What Is an Artifact?
Digital artifact can be of any content types including text, audio, video, image, animation or a combination. [Wikipedia]
Every screen shot on your phone is an artifact. Every show on any screen in your home, every TikTok, every video on YouTube, every image from Pornhub or My Modern Met. All artifacts. Wolf Blitzer on CNN. Your Zoom call with Doctor Orthopod. Artifacts. Now virtually every moment of our days and lives is spent interacting with artifacts. (Just because we have adjusted to seeing and interacting with the world via screens, does not make this situation less artifactual.) We have so adapted to the reformatting of reality that we no longer realize the format-hijack.
Capitalizing on this adaptation are mirror worlds, the emerging reality that makes artifacts so important. Effectively, a mirror world is a mirror of any environment that is reproduced digitally to inform or regulate a real-life situation. Imagine, for example, a busy intersection where there have been numerous accidents when people were seriously injured or killed. A given municipality then sets up a mirror world, with cameras observing the intersection from various angles to monitor traffic patterns and determine how to slow down the rate of accidents. The mirror world shows the real world, more clearly and definitively over time than any single memory or even a jury box of witnesses. Yet, obviously, this mirror world is not the world, it is an artifact made from the real world. An artifact where motion can be stopped, reversed, sped up, blown up; a license or face can be tracked and identified. In short, reality in the artifact can be manipulated in ways the real world cannot.
Now multiply that single example environment by vast swaths of so-called reality. This is not hard to imagine because China has done so already during the Coronavirus pandemic. They were able to mitigate the effects of lack of compliance to social distancing measures by creating a mirror world of their entire country. Using “social credits” and other measures, they were able to constrain their citizenry and keep them from spreading the virus. Each citizen effectively became an artifact:
The privacy implications and alarms regarding this level of surveillance are too numerous to address here — but they must be monitored and addressed continuously.
You as an Artifactory
Fact: a thing that is known or proved to be true.
Artifact: an object made by human being, typically an item of cultural or historical interest
Replica: an exact copy or model of something, especially on a smaller scale
These three facets of previously discreet reality are melting into one another. While you were busy providing essential data to doctors, lawyers, banks, or merchants, you have deconstructed. This is the nature of digital technology which deconstructs everything to miscellaneous artifacts. Digital technologies run the world. (The logical next step after Marc Andreesen pronounced that software eats the world.) You, deconstructed, are being used, and often sold off, for pieces and parts: you are an artifactory, a veritable production engine of data that gets turned into artifacts. As Tim Cook, CEO of Apple, described the personal data capture process recently: “The end result of all of this is that you are no longer the customer, you are the product.”
The steps are simple.
1. Fact (you)
2. Digitized (aspect of or fact about you)
3. Deconstructed (this aspect of or fact about you separates from you, as in the digital determination of your hemoglobin level from a blood test on a specific date)
4. Formatted (this hemoglobin level enters software as a data point)
5. Multiplied (deconstructed and formatted) this artifact — unlike the physical you from which the artifact was generated — can now be replicated, duplicated, multiplied and, of course, used in various ways
6. Artifact is removed from its original context (you, your physical self) and is ready to take on its next life
- Artifact enhances fact (helps the doctor understand your blood chemistry)
- Artifact distorts/fakes fact (the possibilities are endless)
7. Artifact operates with a different logic than physical you or previous human tools such as the alphabet
That might seem abstract until you take a look at what we all have become:
Artifacts we can see and identify are one emerging reality. Another equally important, and potentially more valuable new technology is the arrival of deliberately camouflaged artifacts. While not readily seen (or understood) by the general public, federated learning** uses a model — a digital artifact — to create a “replaceable truth” or said differently, uses an artifact of the data to create a secure, private database. While the details of federated learning’s utility are still evolving, most relevant to this discussion is the artifactuality of this process. Researchers such as Ramesh Raskar start with facts, with patient data and then create an artifact of the data to manipulate some aspect of that data. In the case of using federated learning in a medical environment, it is to secure privacy.
Considering process, federated learning is similar to image manipulation used in image fakes. You start with the real thing — hospital patient data, an individual’s face — and manipulate that data or image to create an artifact of the thing. This is, as Michael Lewis termed it, a new new thing. This is territory we have not visited before. Yes, we have captured the real world, first in words, then images which are themselves artifacts; but we have not been able, until now, to employ the artifact as a changed entity to change the world.
To date, an artifact has not readily lived an existence deliberately separated from the fact. In the case of federated learning, the artifact is used in some measure to hide the fact (for the sake of ensuring privacy). This artifact afterlife is completely unique in human experience.
This afterlife expresses a new dynamic: the enlivening of information.
At first, this may seem like yet more eye-entrancing digital motion that captivates us on our screens and devices. But once we extrapolate how an artifact with a separate life and logic departs from the fact that spawned it, it dawns that artifacts constitute a completely new reality with a new logic, and especially a new destiny.
Are artifacts a new life form? Evidence supporting that suspicion: the more life a fact has, the more it is artifacted, the greater likelihood it can impersonate. An artifact can appear to be a real entity such as a person, a government, a government agency, a doctor, etc. An artifact evolved from a fact is ripe for enlivening. What then does enlivening mean? First, we do not fully know. How we will use and exploit the full utility of artifacts is still in its infancy. Moreover, the technologies associated with artifacts are advancing apace and we are finding more and more useful applications for artifacts.
We might start an exploration of enlivening with a foray into the emerging logic and dimensions of artifacts:
- Once an artifact has evolved (devolved) from a fact it has the ability (the digital option) to separate from the logic of the fact that spawned it
- While facts follow a linear, alphabetic logic, artifactual logic is digital logic: miscellaneous, malleable; digital logic is not stable but able to morph or be morphed
- As a result of the variability of digital logic, the identity, or identifying characteristics of an artifact are not fixed
- An artifact can have multiple, even innumerable, identities
- New technologies (AI, deep fakes, etc.) can enliven the artifact with voice, gestures, facial movements, exact sounding and seeming expressions
- A digital artifact can alter essential characteristics in small and large ways: i.e., an image of a face could change genders or races
- Artifacts can be used to verify facts, as a driver’s license is used to verify identity — yet their reproducibility can make such verification suspect
- Artifacts can be built from facts that use (a model based on) those facts to enhance understanding and utility of the data while shielding certain identifying characteristics of the data. This has remarkable utility for securing privacy, while also raising concerns of such technologies falling into the hands of those who would seek to distort the facts for malign purposes
- Data on user response to artifacts is scant at the moment, but there is a strong likelihood that users will engage with artifacts — especially those that adopt or mimic human characteristics — as though they were human
Artifacts | Bill of Integrities
What are we to make of this new digital twin living amongst us? My suggestion, which some may find radical, is to anoint artifacts with human-like powers — they certainly have, or will have that — and to assign to this emerging class of near-humanity a moral code, a behavioral framework; effectively, a Bill of Integrities.
My first draft of that Bill of Integrities would include:
Integrity of Speech | An artifact has the right to free expression as long as what it says is factually true and is not a distortion of the truth
Integrity of Identity | An artifact must be, without equivocation, who or what it says it is. If an artifact is a new entity, it can identify accordingly; but pretense to an existing identity (other than itself) is a violation of identity sanctity
Integrity of Transparency | An artifact must clearly present who it is and with whom, if anyone, it is associated
Integrity of Privacy | Any artifact associated with a human must protect the privacy of the human with whom the artifact is associated and must gain the consent of the human if the artifact is shared
Integrity of Life | An artifact which purports to extend the life of a deceased (human) individual after the death of that individual must faithfully and accurately use the words and thoughts of the deceased to maintain a digital presence for the deceased — without inventing or distorting the spirit or intent of the deceased
Integrity of Exceptions | Exceptions to the above Integrities may be granted to those using satire or art as free expression, providing that art or satire is not degraded for political or deceptive use
Artifacts may stand-in for human aspects, but they cannot become — we must not let them become — stand-ins for our humanity. To ensure this outcome we will have to think differently about ourselves in a world where technology can deconstruct any entity to its components and then reorder miscellany to a new agenda. We would be wise to monitor that agenda carefully. Artifacts are a new man-made life form, made from us.
*In information science, an artifact is any undesired or unintended alteration in data introduced in a digital process by an involved technique and/or technology. I am not using artifact in this sense; rather, my intent here is to bring to light our burgeoning ability to create digital entities from facts and things.
** “Research on diagnostic uses of AI has stayed narrow in scope and applicability. You can’t deploy a breast cancer detection model around the world when it’s only been trained on a few thousand patients from the same hospital.
All this could change with federated learning. The technique can train a model using data stored at multiple different hospitals without that data ever leaving a hospital’s premises or touching a tech company’s servers. It does this by first training separate models at each hospital with the local data available and then sending those models to a central server to be combined into a master model. As each hospital acquires more data over time, it can download the latest master model, update it with the new data, and send it back to the central server. Throughout the process, raw data is never exchanged — only the models, which cannot be reverse-engineered to reveal that data.
There are some challenges to federated learning. For one, combining separate models risks creating a master model that’s actually worse than each of its parts. Researchers are now working on refining existing techniques to make sure that doesn’t happen, says Raskar. For another, federated learning requires every hospital to have the infrastructure and personnel capabilities for training machine-learning models. There’s also friction in standardizing data collection across all hospitals. But these challenges aren’t insurmountable, says Raskar: “More work needs to be done, but it’s mostly Band-Aid work.” From MIT Technology Review, “A little-known AI method can train on your health data without threatening your privacy,” https://bit.ly/3dEZWlb