"Imagination is more important than knowledge. For knowledge is limited, whereas imagination embraces the entire world, stimulating progress, giving birth to evolution."
-Albert Einstein
Recently I've been making one deep dive after another into cloud and networking technologies, and although this would bore most readers, the available AI tools have lead me to some innovative ideas. Personality simulators are something that's fascinated me for a while now, and perhaps it's just the second season of West World that's inspiring me, but I feel like this is something we should talk about.
The basic idea is that we could create a 'chatbot' which is so sophisticated that it is capable of much more than just superficially plausible conversation and can actually simulate the basic cognitive processes and personality of an individual, to the point that they could stand in as a proxy for the actual person. Depending on the nature of the person we're talking about, this could represent an impressive feat. We would have effectively cloned an individual's personality in an online simulation, allowing us to interact with that individual in a controlled virtual environment without the inconvenience of requiring the individual's presence. This idea has been around for a while, but big data technologies provide powerful tools for data aggregation and pattern recognition which make this more plausible than it has been up until fairly recently.
Ignore Siri and Bixby for now; imagine that the technology will eventually take care of itself. Once we can get this to work, this would be an extremely powerful technology with far-reaching implications. As with most powerful new technologies, these bots would have both powerful upsides and downsides, and we do owe it to ourselves to discuss both. Some possible examples would include:
- Use natural language processing to develop a CharlesDickens-bot, which could churn out new works with all the same plot twists and turns, wit and wisdom as the original author
- Create an Einstein-bot to come up with new theories, making use of the imagination, which he acknowledged was so important.
- Create a bot version of a parent or loved one for a child who is unable to see them in person (i.e., overseas vets or even parents who die from incurable diseases).
- Create a bot version of business associates or friends to provide a practice version of interactions prior to actual negotiations.
- And similarly, here is the inevitable downside:
Already the negatives are staring to appear even with these basically positive examples. The main problems would appear to be:
- Identity - if bots become so effective at simulating real individuals, they could be used to convince others of their actual identity or be used to obtain secret information about them.
- IP - bots with the ability to originate work could be capable of originating work that is similar or indistinguishable from that of the original owner. For example, if I could clone a living author like, say, Tom Clancy and generate new works by him, then those works would arguably take away from the readership which is rightfully his.
- Autonomy - we like to think of ourselves as unique and irreplaceable individuals. The thought that we could in any way be substituted by a bot is inherently threatening to us. As with previous technologies (outsourcing, automation), bots could have a very real and negative impact on individual incomes as well.
"Nevertheless, not to extinguish our free will, I hold it to be true that Fortune is the arbiter of one-half of our actions, but that she still leaves us to direct the other half, or perhaps a little less."
--Machiavelli, The Prince
Our actions are inherently non-deterministic. Even if we were to bootstrap an individual's personality, any given decision point would contain a large amount of unpredictability (be it randomness or what M calls free will). So it's possible that we could have accurate simulations of clearly defined decision points (the yes/nos), but for most of life's small decisions there would be too little information to effectively predict them. In the end, this seems to be one more revolution that is on its way, and our best course is to come up with the means to deal with and perhaps benefit from it, rather than sidestepping it.