Market failure, Pace layers and AI
Remixing principles of market failure with pace layers to understand the AI disruption taking place.
WIP thoughts on AI disrupting the system
Free markets and public choice theory are very much part of current zeitgeist, even if technologists are not talking about it. Advent of AI has clearly disrupted the economy. But the trillion dollar question is whether there is a market failure that deserves some form of intervention( not only government but also organisational).
Before we get into the impact of AI on the overall job market, we first learn what is a market failure and how to solve it from Ajay Shah, Solving Market Failures.
The standard checklist of market failures is:
- Externalities -- e.g. a factory that pollutes.
- Asymmetric information -- e.g. safety in food or medicines.
- Market power -- e.g. firms that earn super-normal profit owing to weak competition, and
- Public goods -- e.g. law and order.
He is writing that piece in counter to the a post by Alex Tabarok and Tyler Cowen which postulates that due to information being accessible more broadly, market failure will reduce. Ajay doesn’t agree and replies with this sentiment.
There is vast asymmetric information in the relationship between an employer and an employee in most complex work places, and nothing has changed which will make a dent to this.
The above statement has been true and continues to be true in the current age of AI. The post was written in 2015. Yet, it feels like there is a market failure in the whole model of organisation setup. People are recommending drastic interventions to take place.
I don’t agree and I would like to walk through the checklist to make my case.
There is an immediate externality that far fewer people are needed to do the same job, especially in software development industry. But it doesn’t mean more economic output is not possible. There is heavy turmoil as we go through the transition but companies are starting up to address many unaddressable problems due to economics of the AI.
We could attribute the AI slop as the asymmetric information part of the product, output from the models is not clear. Also, very few are able to harness the technology to its full potential. But it is more a function of AI’s jaggedness and tools that are currently used to harness rather than the model’s capabilities.
The AI companies market valuations are sky high but there is no monopolisation play here. The open source models are not too far behind. Previous generation business models, like SaaS, are loosing their margins due to the advent of AI. Thus, more companies are competing with one another for the same wallet share.
The implications of this technology has not drastically disturbed the society’s fabric like social media. Because AI is fundamentally a production technology that distributes capabilities of a few to many unlike Social media that centralises.
Now, that I have presented a case of no market failure, it only makes sense to understand that pace of change causing the discomfort in an organisaiton.
There is no better way to revisit it than Stewart Brand’s Pace layer’s framework.
The order of a healthy civilization. The fast layers innovate; the slow layers stabilize. The whole combines learning with continuity.
The layers are in reverse when overlayed with an Org adopting and building with AI :
- Nature - Job market
- Culture - Organisational culture and leadership.
- Governance - Team Structure
- Infrastructure - AI and digital tooling around it
- Commerce - Business models.
- Fashion - Products being offered to customer
Julian Stodd who is asking a similar question. He is writing about his year long pursuit to study the future of work in the AI age. In his post on change of System, he writes AI invalidates the system.
Some change occurs within an existing system, delivering efficiency, optimisation, adaptation, incremental change.
Other types of change simply invalidate a system. They make what came before abstract, irrelevant, diminished below the point of effect, or simply occluded by what has emerged.
When considering the human and AI future of work, this may be one of our lenses.
Right now what I witness is bottom layers are also moving fast with faster innovation at the top layer. None of the layers are stabilising and system is not slowing down.
This is leading to those inside organisations and people looking in, feel the uncertainty. In all my years of work, many people I met dislike uncertainty and that is totally a reasonable approach to life. Schumpeter would say otherwise when it comes to entrepreneurs. I believe AI enhances that trait in all of us, whether we think ourselves to be an entrepreneur or not.
They have not accumulated any kind of goods, they have created no original means of production, but have employed existing means of production differently, more appropriately, more advantageously. They have “carried out new combinations.” They are entrepreneurs. And their profit, the surplus, to which no liability corresponds, is an entrepreneurial profit. Excerpt From - The Theory of Economic Development by Joseph A. Schumpeter
I will try to unpack trends that I am witnessing in the future posts. As I work across the layers in an organisation that is pursuing the change under an honest leadership trying to harness the technology in a value generation capability.
Round up
Couple of things came out that were done earlier but only released publicly in the past 2 weeks.
I did a part 2 podcast with Jonah McIntire on AI and org design. Part 1 which kicked off the series is over here. Its a wrap for now unless we are inundated with questions that necessitate an another one.
Part 2 Interview on AI and Org design
The podcast was a long one but we covered a lot of ground. I prepared my chain of thought but didn’t prep with too many questions. Hence, it was more free flowing and we had help on production front as well. Self recommend listening to it. As I said it last time, Jonah’s views on the AI are underrated and luckily we have things coming out built on the same principles he mentions in the call.
Following the interview, Jonah published a blog on our official comms talking about our efforts on AI front available to customers.
A single, well-scoped AI agent does not have these problems. It has the opposite properties.
The organizations seeing the strongest early results from AI in logistics are not the ones that deployed the most agents. They are the ones that deployed one agent, scoped it tightly to the workflows where the cost of friction is highest, and gave it genuine execution authority within clearly defined boundaries. In practice, that means a single agent working seamlessly across key logistics workflows.
One system. One audit trail. One set of decision rules. One place to look when something needs explaining. One team responsible for its behavior. One governance framework to maintain and improve.
These were in the making for a while but happy to see them out in the public. I don’t work on the team that is building our agent, but they are 100x better than yours truly.
Links that resonated
There are many links that I read in the past 2 weeks. So, one for each week.
The prevalence of AI is fundamentally breaking the model and Matt Webb writes how he feels about it. A classic piece based of noticing the environment beyond the surface.
Om Malik who has been an inspiration to many outsiders looking towards the technology space. He was a journalist, critic and an investor but most importantly he was a decent man. He passed away and there was a galore of personal anecdotes of his graciousness towards them. You should read them.
Sign off
Last 2 weeks have been intense, writing down everything I was thinking was hard because I was overly stimulated with all the travel.
I have been reading every published piece about Om which became public while I was in Europe, surrounded by people I work with. It silently shook me that here was a man who was cherished by everyone for what he did for them. How he made them believe in themselves without being a cheerleader, but a concerned person who cared sincerely.
Yet his personal life was tragic. There was this quote from one of the posts, “he was a very private person but not secretive” and that resonated with me. I believe that I am like that and anyone who has known or worked with me has been witness to some of my unfiltered thoughts in private. I appreciate you all.
I think how he wrote his last post was most telling of his nature, taking a few days off when he was in ICU for couple of months by then.
Be useful to fellow humans is the sense I get that Om imbibed the most and I believe that is a noble pursuit.
Signing off till next time,
Vivek, missing german beers
