Who is Neil Lawrence? Or AI and Gardening
Who is Neil Lawrence?
Why do I ask?
The Financial Tims published a letter he wrote in which he directs us to the real value in artificial intelligence. It will not be sourced through some centrally orchestrated “grand plan” that investors in various AI businesses can more readily put a value on. Not at all. The ultimate wroth of AI will come from the efforts of millions upon millions of users spread across diverse occupations and interests, each of whom will have idiosyncratic reasons for their AI use and all of whom will accept or reject its utility based on the particulars of their own situations.
That’s my interpretation of what he said.
I hope he agrees. It’s how I see the future of AI also.
The uptake of a new technology like AI is dependent entirely upon its local utility, not on the public relations and stock price boosting outbursts of speculators associated with Wall Street or, worse, Silicon Valley. Is AI useful? If so, what for?
Probably. We will find out.
My wife loves AI. She is currently developing an app. She is doing so entirely alone and without the usual start-up paraphernalia offered by the speculators who move in to extract rents from innovation. AI has given her the opportunity to go to alone. Truly alone. Un-obligated to the usual extraction crowd. Yes, she began life as an engineer, so using AI has come quite easily to her, and, no, it is not a replacement for her local knowledge — her app has to do with gardening and is based upon our experiences here in Vermont. AI is an enabling technology in her efforts. It is not a replacement.
Let’s shift gears:
Gardening might strike you as an odd place to begin, so let me explain.
Everything is information. Everything. By which I mean that information is one of the fundamental ingredients to all activity, which when combined with the other fundamental ingredient — energy — everything we see around us can come into being. And as we have all been told, information is physical. That is to say for something to be information is has to take a physical form.
Our garden is a prosaic manifestation of this philosophy. All gardening is based in information. So it is relatively straightforward, if tedious, to translate that information into an app, shake it around, and extract utility for others. Why does that plant grow in that particular spot? Information. Why does it have that root system? Information. Why is it that color? Information. Will it grow in your garden? Information. And so on. The value of AI in this case is assembling the vast experience and knowledge previously crowd-sourced by humanity throughout its history of contact with horticulture. AI is, simply put, allowing my wife to have a conversation with all those who have had similar conversations in the past. And it helping her organize that conversation and pass it along, suitably simplified, to others.
Wonderful.
In his letter Professor Lawrence says much the same thing. He raises the notions of locality and decentralized utility extraction. His example are things like teams working on corporate activities such as payroll or tax compliance. He is corresponding , after all, with the FT! But his choice touches on a sore point.
The corporation itself being the sore point.
It is easy to forget that the modern corporation is a technology itself. It is a method for the manipulation of information in order to produce stuff. It’s great advantage is that it puts a wall around a set of activities that need coherence in order to achieve that production. It is a coordinating structure. It substitutes for the free-for-all of the open market by so doing. The open market being an inadequate place for establishing such coherence — automobiles are artifacts of a corporate setting, not the marketplace. This is why, incidentally, economics has such difficulty with understanding real economies — all the hard activities take place before products enter the market for exchange.
I have always argued that better, easier, or cheaper methods for the movement and cohering of information would undermine the need for the expense of the management involved in production. All management is an overhead. It is a cost of coordination. Economists talk of “transaction costs”, they should really be talking about “management costs”. It would help focus them on the greatest ever productivity bonanza available to us: deconstructing corporate America.
The resistance to even the suggestion of this idea is enormous. We already know the extent of the inertia involved. The pandemic gave us a great experiment in decentralization. It was but a baby step. People could be moved from the real estate and urban hubs that are a consequence of industrial organization methods. They could avail themselves of modern communication and technologies. They could combine effectively and create the coherence that a corporation exists to provide, but at a distance, and at lower cost — the real estate was shown to be unnecessary. They could, that is, still have the necessary communication and conversations needed to coordinate things, but they could it in amore amenable and comfortable settings.
How did corporate America react? By refusing to innovate! It seems that it is one thing to innovate products, it is another to innovate more fundamentally. Inertia, and a certain amount of management fright at the potential loss inherent in those obsolete buildings, prevented technology from moving us into a truly post-industrial world. I suspect that AI might, for the same reason, suffer a similar fate.
Or maybe not.
If traditional management mindsets dominate, as they probably will, and as the endless search for more profit plays out, AI is likely, at first, to be deployed as a labor-substituting technology , not as a labor augmenting one. There will be fewer people. So, perhaps, there will be fewer offices after all.
Alternatively we can resort to history for another narrative:
For AI to become what its supporters think it can be, we need a populace able to use it properly. That is a long term goal. It needs us to re-organize our education system to produce AI compliant workers. Just as we did for industrialization when we began mass education to raise literacy and numeracy to the levels needed for mass-production and mass consumption. WE have done this before, in other words, we can do it again.
This doesn’t mean we need more engineers. What we really need is more people who can engage in conversations with AI. We need to understand what AI really is: the accumulation of human thought and debate. Once we humanize AI in this way and enter into conversation with it, we can extract what Professor Lawrence set as its greatest value: answering local questions.
And then, naturally, those local solutions will find their way back into Ai where they are available to add wisdom for someone else to access.
Ai is simply a mechanization of humanity’s past and present knowledge. It is an accumulation of our collective knowledge and intelligence. Its ultimate value resides in the questions we ask and in our willingness to contribute our knowledge to the store for others to access — our use of AI produces information for others to use. And we must remember that, in effect, when we deploy AI we are asking humanity to respond.
This raises, naturally, questions about who owns access to our collective knowledge. That, I think, is the ultimate challenge for democracy — one that is being failed at present. AI is ours. We need to own it. It is the ultimate utility at our service. But that is another conversation, one about the true meaning of democracy.
Meanwhile.
Why not have the wisdom of all the gardeners in the world — ever — to help know whether that plant will grow in that, particular, and very local, place? It is, after all, only information.
