A dear friend sent me on WhatsApp a picture of himself looking strikingly younger, slimmer and well-dressed, standing beside a classic car, writes Mark Rowe.
I was about to reply congratulating him when he sent further photos of himself in poses as a medieval knight. It dawned on me that the pictures were not real. He confirmed in writing that the images were the result of AI that his children’s school had banned. A commentary on both how irresistible AI is and how pointless bans and edicts by anyone in authority are against something so easily available.
In his speech last week on AI to the Royal Society, ahead of the AI Safety Summit at Bletchley Park tomorrow and Thursday, Prime Minister Rishi Sunak likened the coming of artificial intelligence (and terms such as generative AI, or ‘Frontier AI’ are already dividing AI) to the Industrial Revolution (a pleasing connection to make, as Britain led in it), electricity and the internet. A better comparison would be with the printing press. Like AI, it did away with entire job roles, unless monks are still copying chronicles by hand in monasteries.
Print made books, the printed word, more available (just as the internet has sent the value of the word to zero, as anyone can post online, under-cutting the traditional, paid-for print media). The single most powerful book was the Bible. Not only did the printing press make Scripture available more cheaply, it prompted William Tyndale to translate the Bible into English. Note how the piece of tech, the printing press, led to the wider change. Why would anyone bother to translate the Bible into everyday language instead of Latin, if only a handful of copies would ever be made. Note how subversive the Bible in language that anyone could understand was. The authorities were correct to regard Tyndale as subversive; he went into exile and was in the end executed as a heretic. The subversion was that if anyone could read the Bible, grasp its meaning, and act upon it, where was the need for priests? That connects Martin Luther, and 450 years of modern history, with Rev Dr Martin Luther King, for the Pilgrim Fathers sailed for America in 1620 as the only way, they felt, to live the religious life they sought.
While many other things caused the modern world besides the printing press, the printed word had consequences, that had further consequences. That would suggest that the approach to AI in terms of safety (‘the future of AI is safe AI’) set out by Rishi Sunak to the Royal Society – ‘where we collaborate with partners to ensure AI systems are safe before they are released’ – is mere wishful thinking. The effects happen after release; effects on society, the economy, and culture, each individual, that we cannot imagine.
To give only one story. In the year 1800, a 15-year-old Welsh girl named Mary Jones walked 25 miles barefoot across Wales to Bala, to buy a copy of the Bible in Welsh. That example of the power of wanting your own copy of the Bible inspired the founding of the Bible Society.
What does this have to do with security management? Such large changes to humanity affect security management and any other occupation. Some AI will be of more direct interest than others. ‘Good’ AI may help spot software vulnerabilities; hackers may use ‘bad’ AI to exploit those vulnerabilities. What we already know is that the scale, and pace, of use of tech will be far greater than ever; implying that businesses, policy-makers and the regulators they appoint, will have to work at pace also; because criminals will work at their own pace.
The omens are not good. Public policy appears to discount security. Take a non-technical example, of the proposed cuts to staffed railway station ticket offices. Either more stations will be unstaffed, or ticket-sellers will roam. Either way, what of the safety of passengers, and indeed remaining staff, if fewer uniformed authority figures are around, to informally enforce norms – a commonplace of crime prevention theory?
Understandably policy around data is in the air, not only because of AI, but while the UK is seeking its own way in terms of data protection since it parted from the European Union and the GDPR. That has had a bearing on the folding of the Office of the Surveillance Camera Commissioner. All the standards for use of video surveillance images, made largely thanks to the goodwill of industry volunteers, will presumably wither. The lesson is that standards for legal or ethical use of data do not necessarily progress; they can be set back.
Arguably the most profound effect of AI will be on the internet, resulting from mass use of AI by my friend (if not his children – in their school, at least). A Government document published ahead of the Bletchley Park event on ‘Future risks of Frontier AI’ mentioned the possibility, under a ‘wild west’ scenario; that the ‘internet is seen as increasingly polluted, with concern about the historical record’. A speaker from the UK official National Cyber Security Centre (NCSC) raised this at the recent Consec, the annual conference of the Association of Security Consultants; that ‘deepfakes’ in a word, synthetic media, manipulated or entirely false video or imagery may be used maliciously, whether to scam people or undermine western democracies. The NCSC man suggested that part of the answer would lie in education of the young …. in schools, presumably.
Photo by Mark Rowe; on-street graffiti, Manchester city centre.
