Risks and Rewards of Artificial Intelligence - BSQ Research

Risks and Rewards of Artificial Intelligence

Revolutionary new technologies like Artificial Intelligence bring with them large scale social and economic dislocations, and often arouse the anxiety that humans are going to lose their place in the world and be replaced by machines. However, history suggests that the usual dynamic of such revolutions is that they start out by replacing human workers by machines, and the development of brand new products and services in the long run generates many more jobs that were lost earlier. Although this optimistic scenario is not unlikely, there is some disturbing speculation that this time may be different, and the impact of AI on society may be more catastrophic than any past technological revolution.

For an example from the past, consider the invention of the heat engine, and the underlying science of thermodynamics, that completely changed the world in the 18th and 19th centuries. The origins of this invention lie in the principle of conservation of energy and the possibility of transformation between different forms of energy, especially of heat into mechanical work. These theoretical considerations led to the development of practical methods for transforming heat produced by burning a fuel into useful mechanical energy. The first version of this technology was the coal fired steam engine, where the fuel was burnt in an external boiler, and the heat used to power, among other things, steamships and railroad trains. Later came the petrol fueled internal combustion engine, that is still in widespread use. The use of fossil fuels to generate mechanical, and later electrical energy, initially replaced the work of people and animals in industry and transportation, but in the long run it led to hitherto unimaginable feats of power generation and usage that were only possible with this new technology. This transformation ushered in the industrial revolution and the modern world as we know it.

The current advent of machine learning and artificial intelligence, and the information, communication, and computational sciences that underlie these technologies, will probably change the world in just as profound a manner as the invention of steam power. This new industrial revolution will liberate people from the need to do repetitive and dull cognitive tasks, such as routine business communication, record keeping and data analysis. In time, it will also take over more complex tasks such as predictive maintenance, autonomous transport, and health care surveillance. In the very long term, one can imagine AI entities taking over a large part of the cognitive work load of a post-industrial society, leaving only the most innovative and creative tasks for humans to do.

Although the current transformation, just like the original industrial revolution, started with the automation of routine tasks, we are already seeing the next stage of the transformation, i.e. the use of information technology to accomplish tasks that were not even conceivable earlier. For example, consider the use of machine learning by banks to detect credit card fraud and trigger real time remedial measures. All banks use this technology that is based on detecting signature patterns associated with fraud. These fraud detection systems process billions of transactions and can respond in real time and send out an alert, and trigger a human intervention. These systems have saved banks billions of dollars in losses from fraud, and have become an essential tool in the technology kit of a transactions processor. The techniques are enabled by ICT technologies that can capture and handle very large data sets, and by machine learning tools that can analyse these patterns and features quickly. Real time fraud detection is a completely new capability, and cannot be considered “task automation”, since such a task was not even conceivable without large scale machine learning and data capture.

The dynamic of starting with the replacement of existing products and services, and continuing on to the creation of completely new products and capabilities, is common to all revolutionary technologies such as heat engines, electricity, and AI. For instance, the first major application of electricity was for illumination, where it replaced gas and other lamps. However, today we use electricity in thousands of ways that no one could anticipate two centuries ago. The initial phase, when technology starts to replace human workers, often triggers acute social anxiety that the economic structure will not survive this disruption and this may lead to a strong social reaction, antipathy, and opposition to the changes. This happened during the Industrial Revolution, when widespread job losses from the use of machinery gave rise to the Luddite movement, a 19th Century worker movement against automation and machinery. Eventually however, the new technology usually creates completely new tasks and opportunities, and the resulting economic growth creates many more jobs than were lost in the initial stages. The Luddite movement faded out when it became evident that the new industrialised world would have more than enough jobs for its citizens. The effect of Artificial Intelligence may follow a similar pattern, and initial job losses may be more than made up later by job gains from the development of new AI based products and services.

It has been speculated, however, that this optimistic scenario will not hold, and that the progress of AI will displace human workers permanently. The reason that this time will be different is that AI represents a very different kind of technological revolution that anything in the past. The difference is that all past technologies have replaced human muscle power, but AI will replace human brain power. Although this will start with routine and repetitive tasks, it may not be long before more complex analytical tasks are taken over by machines, and sometime in the future it is possible that even highly complex cognitive tasks will become the domain of artificial intelligence. This will leave people in an AI world with little to do.

Another source of risk comes from an activity at which humans excel, the collective hunting and killing of other animals, including other people. Excellence in this area has placed humans at the top of the planet’s food chain. The development of artificial intelligence and weaponised robots means that our position as top predator on this planet is in jeopardy.

The loss of our position as the best thinkers, and the best killers, on the planet may turn out to be a catastrophic development for humankind. Some important thought leaders, like Stephen Hawking and Elon Musk, have warned that this could turn out to be an existential risk for humanity. Others, like Mark Zuckerberg, have maintained that such fears are overblown. AI researchers have pointed out that although such risks do exist, they are far in the future. General artificial intelligence and terminator robots only exist as fiction now and this is likely to be the case for the rest of this century. Any forecast of these developments is very speculative and likely to miss. What we do know, however, is that in the shorter run AI is going to be deployed extensively in our homes, offices, factories, and roads. It is going to throw a lot of people out of work, and it is also going to create a lot of new jobs, and we will need to make a collective effort to smoothly incorporate these revolutionary changes into our economies and our societies.

Leave a Comment

Your email address will not be published. Required fields are marked *