Mythbusting AI and ML in Local Services
Melvin Kranzberg is best known for his “Laws of Technology” - intended as truisms based on his experience of how technological changes impacted society. Of these, the most famous and prescient is “Technology is neither good nor bad; nor is it neutral.”
In offices and conference venues around the country terms like Artificial Intelligence (AI) and Machine Learning (ML) have proliferated. Concepts which were once the domain of human/ robot chess matches and academic papers are now part of our everyday lives.
However, despite their prevalence in conversations, misconceptions abound, particularly in how they can be leveraged for the public good. On one hand, advocates see a silver bullet - here to radically improve every part of decision-making and implementation. Cynics see the entrenchments of human biases and inequalities without the moral oversight or human nuance required for making good decisions. In reality, both sides make valid points, but both are also often missing the actual value that AI/ML can add.
Before we get into the detail, it is worth looking at what AL and ML actually mean, as they are often used interchangeably. They are similar fields, but this isn’t quite right.
Put simply, Artificial Intelligence is a concept - effectively that machines be made capable of doing things which previously relied on human intelligence. Currently much work is focussed on the interpretation of senses and communication - understanding sight (machine vision), written/ spoken languarge (natural language processing or NLP) and thought (knowledge representation and reasoning).
Machine Learning is part of the field of AI. Essentially it is the process through which machines improve their performance without the requirement for a programmer specifically telling them to do so.
A good example of this is the evolution of spam email filtering. Originally, a person would have to fully manage their own inboxes - categorising or manually removing spam. Over time, developers built rules to recognise certain indicators of spam (usually blacklisted senders) to automatically intercept it. More recently, algorithms have been trained on datasets of emails that are pre-labeled as spam or not spam. This training allows the model to learn and recognise the patterns and features associated with spam to be able to intercept it far more effectively. Training data can be continually updated to address new tactics/ threats, however the most advanced approaches (linked to NLP) look to learn in real time as emails are received to protect against spam.
Over the past few weeks I have spent time considering my own experiences, as well as others working in innovation and AI in and around the Public Sector. I have come across a number of illustrative examples of misconceptions which are worth further discussion:
- “Data is the New Gold” - Is data inherently valuable, or does its value depend on usage and context?
- “AI is Only for Tech Companies and Requires Large Teams of Data Scientists” - Can non-tech organisations leverage AI without extensive expertise?
- “AI isn’t Suitable for Policymaking as the Real World is Too Complicated” - Does AI have a place in complex decision-making environments like policymaking?
- “AI Enables Us to Fully Automate Services and Drastically Reduce Costs” - Can AI really lead to complete automation and significant cost reduction in public services?
I’ll go into these in more detail in my next blog post, but if you have any other suggestions of myths (positive or negative!) in the meantime please do let me know in the comments below and I will look to address those as well. Please do reach out to me or to techUK if you want to know more about AI!
Local Public Services updates
Sign-up to get the latest updates and opportunities from our Local Public Services programme.