Back to the 70s: The awesome power of AI
PDMS Founder Chris Gledhill shares his first encounter with AI all the way back in the 1970s and how it's changed today
I will never forget my first encounter with the awesome power of artificial intelligence! It was a dull September day in 1974 and there was a real buzz in the playground. A queue had formed, as most of Year 9 waited eagerly for a go with the first electronic calculator any of us had ever seen. We each took our turn to marvel at the unbelievable sums this incredible (and surprisingly affordable) machine could perform in an instant.
Predictably, the entire educational establishment had a nervous breakdown, worrying that this technology would breed a new generation of mathematical dullards who were unable to read a set of log tables. Meanwhile, we rapidly moved on from pointless hard sums to the far more creative pastime of spelling out rude words on a rudimentary LED display –5318008 anyone?
Fast forward 50 years and here we are again, metaphorically at least, queueing up in the playground to marvel at the incredible power of generative AI. According to Moore’s law - wait let me find my calculator - the computing power we have available to us has increased by 2 to the power of 34, or about 17 billion in round numbers, since those innocent days in the playground. Now, we can use the collective power of the world’s computing resources to instantly write an essay on the power of AI in the style of Donald Trump!
Joking aside, the current state of AI, like the electronic calculator, the steam engine or electron microscope, is a truly awesome achievement. Our ability to build tools which extend our abilities - to travel at speed, lift heavy loads, understand the world beyond the limitations of our own senses and so on - is what really makes us human.
Put to an appropriate use, these new digital tools can help us to solve some truly challenging problems. Back then, we soon discovered that the electronic calculator didn’t actually make anyone better at maths if they didn’t understand which calculations were needed to solve a problem. Similarly, AI, despite its flashy grammar and faultless spelling, doesn’t actually know anything.
What is does incredibly well, however, is process vast amounts of information to reach a defined goal. Like all transformational technologies that we’ve come to accept as part of everyday life, AI makes some tasks that would be time-consuming and laborious or even impossible for us, seem easy. But what it doesn’t do is know whether the task was worth doing in the first place.
Success stories in the application of AI, such as improvements in medical treatment through more accurate interpretation of diagnostic imaging for example, generally follow a similar pattern. A system is trained using existing information to see patterns in data (images in this case) which even the most skilled human would probably miss, then flag up the possibility that this is associated with a defined medical problem. The skilled human can then use their judgement to determine what is an appropriate next step for their patient. The machine amplifies the doctor’s ability to process images quickly and accurately, freeing up scarce human resources for the more cognitively demanding tasks that may arise from a diagnosis.
Put more generally, the tool is trained to process specifically defined information more quickly and accurately than we can, freeing up the skilled human to be more effective at what they are trained to do. Deciding what information to process and what to do with the results is and always will be the responsibility of the people using the tools. AI, like any other tool, will be incredibly useful in the hands of skilled people, but at best, a waste of time and energy if used without careful and targeted design.
Unfortunately, one of the perverse consequences of the launch of general-purpose generative AI services like ChatGPT or CoPilot has been a slowdown in investment in more conventional IT investments in some sectors. Understandably, decision makers are hesitant to spend tight budgets on mere incremental improvements in productivity when there might be something truly transformational available just around the corner - if only we knew what it was. In reality, most organisations have much groundwork to do before they are even in a position to benefit from the widespread adoption of AI within their primary functions.
The good news, however, is that most of these preparatory tasks come with their own, often quite significant benefits in terms of productivity, customer service, and organisational efficiency. The key to success in the adoption of IT tools of any generation has always been the same: do you know who is going to benefit from your investment and how it serves the purpose of your organisation? In other words, how are you hoping to use the tools of AI to amplify the impact of the people you employ, and better serve the people you supply - be they customers, citizens, patients, residents, pupils, collaborators or anyone else.
And there we come full circle. Just as we once marvelled at the power of a simple calculator to revolutionise our approach to maths, today we stand on the brink of another technological leap with AI – wondering where it will take us and how we can best harness its potential.
To find out more about AI and how PDMS can support you, visit our AI & Data page