The impact on infrastructure once artificial intelligence shifts into top gear
23 September 2019
Standard computer programming has been around for decades; what is new (ish) is the propensity for computers to teach themselves how to spot patterns and progressively improve.
Arup’s Tim Chapman points the way forward for people planning a career, which is likely to last at least 45 years, and through which they will encounter unfathomable change.
It is very easy to become fearful when reading press report after press report about the impact of artificial intelligence (AI) on our civilisation – all the jobs to be lost and what will become of us? Or our children?
First, it is worthwhile being clear about what AI actually is. Definitions can all too easily become conflated with the latest scary film – portraying robots with high intelligence and even human emotions – great films like ‘I Robot’ and ‘Ex Machina’ are wonderful stories and do show what may ultimately happen – but the level of technology is many decades away, if it ever happens.
What is more insidious is the progressive augmentation we get from ever more adept systems. Standard computer programming has been around for decades; what is new (ish) is the propensity for computers to teach themselves how to spot patterns and progressively improve.
And generally these algorithms are good for us, such as the ones that learn how to detect potentially cancerous moles on skin – initially learning from the best doctors – and thereby becoming better than any of them, then learning from historic photos with a precise diagnosis of which ones did actually turn cancerous later.
This is an application of machine learning or artificial intelligence in action – the whizzy clever robots which can do anything are called artificial general intelligence (AGI).
So then what is intelligence? Einstein wrote, “The true sign of intelligence is not knowledge but imagination”, while Socrates wrote, “I know that I am intelligent, because I know that I know nothing”. It is unlikely that your HP laptop is anywhere near any thoughts that profound.
Nine types of human intelligence:
4.) Logical-mathematical –understand principles of a causal system
5.) Bodily kinaesthetic – sports, dance, acting, making things
6.) Interpersonal – social skills
7.) Intrapersonal – self-reflection and aware
8.) Naturalistic – nurturing information to natural surroundings
9.) Existential – spiritual
Currently, AI is making some progress in only a small portion of these areas, fortunately. In the field of original composition, AI is making some progress in art and music writing, but mainly by averaging many prior human art.
Cloudpainter won the 2018 Robot Art Prize with a decidedly confused pastiche and ‘the Portrait of Edmond Belamy’ was exhibited at Christies with an asking price of £7,000 to £9,000 – it was made of an amalgamation of 15,000 portraits from the 14th to the 20th centuries – so is far from original.
Every year the firm Gartner come up with its ‘hype curve’ for new technologies – plotting the progress of each from an innovation trigger through a ‘peak of inflated expectations’ towards a ‘trough of disillusionment’, eventually into a ‘slope of enlightenment’ and hopefully reaching a ‘plateau of productivity’.
Various AI technologies can be found throughout all of these zones, with AGI at the most undeveloped end.
A context of world trends
It is worth putting AI into a context of world trends, which can combine to either thwart or reinforce existential threats – so AI can be seen as either a saviour or a reinforcer for risks to humanity’s future like global warming, resource depletion, destruction of our environment and deteriorating global order, alongside more usual threats like disease pandemics, which we thought we had cured but antibiotic resistance could allow to return.
Another interesting facet of this trend towards computer assisted process improvement and ever more expert systems is where does it leave the current human experts?
The professions derive their exalted position in society from the pact made at the time of the medieval guilds and it has been unchallenged until now.
Now various professions are being dumbed down by the invasion of expert systems, initially amplifying and improving expert opinions, but eventually supplanting them, apart from a small number of more complex cases.
This could easily lead to a reduction in status and salary for adherents to those professions. The recent 737 Max crashes illustrate the perils of uncontrolled trust in AI systems, but also show the zeal with which such systems are intruding into activities that we consider to be human controlled. Will truck and train drivers be needed in the long term?
Overturning standard business models
AI systems can also disrupt industries in other ways by overturning standard business models. Hence Uber is the world’s biggest taxi company but owns no taxis.
Facebook is the world’s biggest media content provider but provides none of the content itself – it is just a platform. Many industries are ripe for revolution in ways we can’t yet imagine.
And these changes now happen quickly. It used to be that a disrupted industry had time to react to change – but now it can occur in months.
This backdrop can be applied to any industry – including that for infrastructure provision. In parallel, we are getting sharper about how we provide infrastructure nowadays. It is no longer the domain of nerdish engineers working in a vacuum plotting lines on maps with less consideration for the communities that will host it than they should have done.
We are much more aware of the special needs of the society for whom we provide infrastructure and which will pay for it though taxes or user charges.
We recognise that it is the outcome from the infrastructure that really matters rather than the asset themselves – and we also know that the successful operation of assets is as least as noble an activity as designing new ones.
Expert systems starting to obviate need for high technical skills
AI is intruding into all of these worlds too, and in some ways the expert systems are starting to obviate the need for high technical skills.
Equally data analytics on users of infrastructure are providing us with fascinating insights about how it can work – and enable us to use these tools to design much better infrastructure that is ever more useful to the communities whose standard of living depends on its successful operation.
It is interesting to muse about whether there will be limits to the levels of intrusiveness that computers will be allowed to reach in our society. While they have the power to render many services quicker and thereby cheaper, what will happen to all the displaced humans?
Initially those that are at most risk of being displaced from the workforce are those with the lowest skills – drivers being an obvious example – what will the people who currently drive taxis and trucks do if that opening is no longer available to them – will there be other jobs that allow them to support their families?
Presently, it seems that governments are becoming weaker and are less capable of taming the concerted global actions of the big tech organisations.
Higher level of control
And the ambition of those large corporations to impose new technologies on us, making our lives potentially easier, but all the time minimising the tax bills that sustain our society and enable us to make our civilisation generous to everybody. Will governments eventually exert a higher level of control or will the big tech firms continue to run wild?
Before we get too worried it is worth reflecting on what computers are good for, and not so good at. We know that they are very very good at:
• Tasks/processes (if programmed well)
• Ordered memory (if designed well)
But not so good at:
• Obscure disordered memory
• Radical rethinking
• Different situations
• Non-routine tasks
Hence an AI expert firm might thrive for two years, but would be incapable of dealing with the changes in our world, not least in the advance of technology.
It is worthwhile reflecting on the various levels at which AI might hit:
• Industry – profound inexorable change
• Firm – inexorable too, with winners and losers
• Person – think of yourself or your children – there will be winners and losers too; therefore we all need to make the right personal choices: staying ahead of the sorts of technology that could make us redundant. This makes it very difficult to plan for a 45-year career
• Society – which depends on how nimble governments are and whether they stay ahead of the global tech firms?
So when AI finally hits the world of infrastructure creation and operation, it is fair to say that:
• Industry will become far more efficient and agile
• And also potentially more responsive to society
• And hopefully less impactful on society – in terms of pollution, which can be more efficiently minimised
• Hopefully making construction cheaper to build – lower user charges, so more affordable
• But blander too
• With fewer people employed
• And fewer experts needed – so fewer peak salaries (some will still be needed though!).
A critical reflection is what happens to those who have no other place to go?
Author: Tim Chapman CEng FICE FIEI FREng, director and leader infrastructure London Group, Arup.https://www.engineersjournal.ie/2019/09/23/52088/https://www.engineersjournal.ie/wp-content/uploads/2019/09/GettyImages-943067032-1024x576.jpghttps://www.engineersjournal.ie/wp-content/uploads/2019/09/GettyImages-943067032-300x300.jpgTechArtificial Intelligence,Arup,machinery