My Recent (& Not Particularly Original) Thoughts on AI

Estimated Reading Time: 7 minutes

We miss things in the AI conversation if we frame AI as a direct replacement for human effort.  The formula is not B equals A.  General and generative artificial intelligence is not going to replace a teacher, therapist, lawyer, social worker, etc in a linear and straightforward fashion. The more that I'm reading about AI and thinking about technology in general as well as looking at how it has played out over the last 20 years, I'm finding that these are the 3 pathways that seem likely.
Lego figurines of R2D2 and C3PO from Star Wars.
"These aren't the bots you're looking for..."
Photo by Mulyadi on Unsplash

. Though not a new idea, this is still a powerful and valuable one.  It's where a specialist with AI support can do more and higher quality work at a faster clip than before.  Individual work begins to scale in this context; when 1 person with a good AI can do the work of 10, 100, or 1000 specialized people. This is a well-established route for technology:  reducing but not replacing skilled labor; deskilling but not destroying an industry.  There are many possibilities for this approach in trade and professional fields.

For instance, folks keep saying that folks like electricians or plumbers or the like cannot be replaced by robots and AI.  But a single plumber or electrician can do a lot more.  They can use bots and AI to diagnose and anticipate the problems of clients.  They may do this through apps that customers download and use to analyze the scene.  Such an app may also be filled with data about the dynamics of the local system that are part of public and online available records. The electrician or plumber can use this information to anticipate the actions, the resources and the order of priority of calls while also using 3D printers and drones to create and deliver the parts that they need. All you need is a company or two to see how they can put this later of technology and automation, and soon, you can easily have more of the work deskilled and automated out.

AI Assistants:  We all begin with AI Assistants trained on our data and interest; they are constantly adjusting and developing in response to our needs and desires.  These Assistants will know us in some ways better than ourselves and anticipate our needs, abilities, and where we need support, suggestions, and help.  They will guide us in what is plausible and possible for us on our own and what is likely needed to be farmed out. This will be similar to social media in that there will be different platforms that offer different types of AI assistance.  Some might be more aggressive; others much more passive.  Some will have personalities and others will just be silent omnipresent entities.   

We're increasingly familiarizing ourselves with these whether we talk to Siri, Alexa, or Google Home. These are the air balloon versions of future jets of AI.  

AI Alternative:  It's going to be a wild ride in this space for a while.  A lot of systems and knowledge of systems are built upon human interaction.  Much of the research that we have about humans and their development is largely structured through social interaction and not necessarily the individual as they exist.  For instance, most of what we know about learning is largely created through learning that is facilitated by and includes another (the instructor). It's also largely looking at learning in confined areas (in a classroom, over the span of a course, or how a few variables impact learning). Nearly all of our research suffers from the streetlight effect--we study the areas that we can see and pay little heed to that which we can't.  

This is true for all socially-oriented professions:  mental health, training, service work, physical therapy, and so on.  Right now, lots of folks are thinking about how n AI bot will replace them in traditional ways.  E.g. Having an instructor that is an AI bot or a weekly meeting with your AI therapist or the social worker visiting the home.  

But that's what we get wrong and where B does not equal A.  B is going to be an entirely different formula.   Instead, life data, embedded tech, or ecosystems will be interwoven to create AI systems that work differently than human service work.  This will take time and lots of trial and error to figure out.  Some platforms may already have an advantage such as wearables and smart devices like the Sleep Number Bed.  But the end result will likely be within a standard deviation of effectiveness and accessibility of traditional human services.

So what might this look like?  Therapy is an interesting one.  What if instead of a series of set-timed dialogues, it happened in a more dynamic way.  First, what if the intake could allow for the individual to unload a lot more data--all communications, writing, texts, etc.  Then what if the intake assessment could be an observation over the course of a week with numerous check-ins & psychological/physiological markings from which an AI might start making recommendations, prompts, or soliciting additional insight. In addition, the AI might schedule times for reflection and conversation that can be further analyzed and added to the case file.  This system might be linked to an actual therapist (i.e. the AI Assistant mentioned above) or human-therapy services or the system may just call on as needed.  

But out of this, there may be a lot more discoveries and understandings--ones that aren't as easy or possible to capture and record given what I mentioned above about how studies in these areas currently happen in limited conditions.  

How or why?  Well, these AI systems will need to be regularly updated and adjusted. Not just the software updates like our phones and computers but also knowledge updates.  The changes in scientific, legal, and other disciplines of knowledge will continue to impact AI and these types of services. As more observational data is collected in correlation with biometrics and actions by large amounts of individuals, it may discover new elements about humans that would not have been understandable or capturable.  

This means there will be more need for access to research (beyond all the internally accumulating research). And I wonder if there is a possibility to legitimize some of this work by requiring such findings and discoveries about humans to become part of the public commons.  That is, companies may not have to reveal their code but they would need to disclose their discoveries related to human psyches.  This could work both to further advance human understanding and hold the companies to a publicly accountable structure (for a variety of ethical considerations).  

The rise of these AI alternatives will likely include different services that provide different levels of involvement and nudging as well as different levels of cost that would also come with different levels of commercialization and product placement. The potential to manipulate or do harm will be significant and also, many people will buy into it either out of curiosity, fear, or feeling the need to do so because of their situation.  

That's exciting, scary, and hella problematic given that there are lots of problems, concerns, challenges, and critiques of large-scale data and machine learning.  Still, I do think this is the road we're going down and so I wonder mostly about what might we be able to discover about humans and how might we also find some controls, guidance, and common structures for protecting people the ways that companies might take advantage of such a massive amount of insight into their customers.

Did you enjoy this read? Let me know your thoughts down below or feel free to browse around and check out some of my other posts!. You might also want to keep up to date with my blog by signing up for them via email.

Creative Commons LicenseThis work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.