Introduction

There is much conversation everywhere at the moment about Artificial Intelligence.  The rise of ChatGPT has awoken a sense of the rise of our robot overloards.  We look to the future for when machines are in control.  I see more of a continuum from business analytics into machine learning and then into artificial intelligence.  Generally, systems that work with language models are more likely to be described as artificial intelligence.  However, I have seen plenty of chatbots that I would not consider to be good examples of artificial intelligence.  But ChatGPT caught me off guard, and made me think that some regulation is in order.  

I want to take this opportunity to look at more obvious systems, where humans and machinery coexist and how humans behave in those systems.  I believe it gives us great insight into the dangers.  The danger is not in the machines.  It is how human beings participate in, and react to those systems, at a societal level.  

Intelligent Systems Defined

One definition of an intelligent system is where a system learns from its environment and changes its actions to better meet its objectives. (1)  It seems a beehive meets that definition. (2)  It seems that an Enterprise Resource Planning system meets that definition. 

The difficulty with Knowing Objectives

It seems that societies are such a system,  although the objective of a society is very hard to fathom.  (3) Maybe it is a manifesto.  But at best a manifesto is a point in time expression of the simplest things that a set of humans can agree on.  The world moves on.  Our collective intentions change.  Our understanding of what our previous intentions were, change.

Setting Objectives and Modifying Behaviour

We codify our collective intentions in a set of laws.  Laws are slow to change.  Laws come with enforcement mechanisms.  Our socialization process, means we are generally reluctant to transgress laws.  We are wired to cooperate.  The degree of machinery in this system may be small.  My areas of examination here is the behaviour of the human beings within the system.  

It seems a company is such a system. (4)  The objectives of a company are in its memorandum and articles of association.  Some might say that those were only ever designed to be limits on its actions and that its objective is profit.  More generous observers may say that its objectives might be in its strategy and key performance indicators.  These, after all, are generally tied to executive compensation.  Investors buy in to a strategy and authorize management to gather the resources to execute on that strategy.   Those resources generally include lots of machinery. 

Measuring the Benefits of Behaviour Change

Profit is an interesting notion in and of itself.  Profit is an economic surplus.  Economic surplus should reflect some notion of an increase in utility to the population. (6) We have to remember that Adam Smith was an ethicist.  He observed that in pursuing an economic surplus for yourself, you could not help but change the market for the betterment of all. (7)

The Unexpected Behaviour Caused by Those Objectives

Companies tend to create incentives and policies that provide some assurance that they will fulfill their objectives.  The human beings within these companies tend to adopt these objectives and enforce these policies.  It changes our behaviour.  

This might not seem a startling observation until you think through the history of what companies have done.  The East India company raised armies, subjugated foreign lands and accounted for half of the world's trade during the mid-1700s. (5)  It makes me wonder whether a man on the street at those time would have a different set of mores.  Would his internal compass be different than mine and to what extent is any difference accounted for in the sets of rules and policies that cascade from these objectives.  It seems that objectives in prior epochs caused behavior that now seems reprehensible.  It makes me wonder to what extent the writers of the objectives would have found the behaviour acceptable.  Did they understand the consequences of those objectives? ,We may think that taking an example from the 1700’s is not relevant to today’s standards, until you remember the ethical purchasing scandals in sporting goods (8) and sewage flowing into UK rivers (9).  

Objectives in a Media System

I want to end with thinking about the humans in machine learning systems and how the participation in AI systems is bound to change us.  Any intelligent system has a set of objectives, but all it processes is an input stream.  In the case of an enterprise that has a business model based on advertising, the objective is high quality ad placement opportunities.   Success against this objective is in part being able to make better predictions from the stream of information about content viewed and interactions with that content.

Where Does the Machinery End and the Input Stream Start?

Another way to fulfill the objective is to make the input stream more predictable.  That input stream is us.  The objective is in part to make us fullfill the objective function, making us less spontaneous, less creative, less human.  This is not just a change of behaviour. Now that we spend between 5 and 6 hours of our time online, our wants are constantly being nudged to make us more predictable.(12)  We are more predictable if we are more extreme (10).  This is a change in our wiring, in our biology. (11)  Changes in behaviour use to be measured in genetic timeframes.  The rise of our species has allowed for changes in behaviour in cultural timeframes.  These systems will learn very quickly how to manipulate us.  The change in our behaviour is now subject to Moore's law. 

Conclusion

We always imagine Dr Frankenstien pulling the level and the spark of life leaping into his creation.  

There was a time when everybody thought man was created in the image of god.  We are headed for a period where man will be made in the image of his own creation. 

References

1. The European Commission’s HIGH-LEVEL EXPERT GROUP ON ARTIFICIAL INTELLIGENCE (2018)

2. The Intelligence Of The Collective. Wildflower Meadows .  The Bee Blog https://wildflowermeadows.com/2019/10/the-intelligence-of-the-collective/

3. How Not To Destroy the World With AI - Stuart Russell https://www.youtube.com/live/ISkAkiAkK7A?feature=share&t=2559

4. The Corporation, Joel Bakan. 2005

5. The Corporation That Changed the World: How the East India Company Shaped the Modern Multinational, Nick Robins 2017

6. The Wealth of Nations, Adam Smith 1776

7. What Is the Invisible Hand in Economics?, Christina Majaski 2023, in Investopia

8. Nike accused of tolerating sweatshops. Burhan Wazir, Guardian 2001 /www.theguardian.com/world/2001/may/20/burhanwazir.theobserver

9. Sewage spills: Water bills set to rise to pay for £10bn upgrade, Esme Stallard, BBC 2023

10. Average daily media use in the United Kingdom (UK) in the 3rd quarter 2022.  Ani Petrosyan, Statista.  

11. How we need to remake the internet, Jaron Lanier 2017

12. Beyond ChatGPT: Stuart Russell on the Risks and Rewards of A.I., Commonwealth Club of California