AI and the election: a voters guide

pupeteer.jpg

An emerging threat to democracy

The UK's National Cyber Security Centre (NCSC) recently raised concerns about the threat of AI to free and fair elections

So what is AI in this context? How does it impact you? Can you defend against it, making yourself less prone to its influence?

NPD

Think of all the people you dislike because they are manipulative. Do you know a person who always seems to know which of your buttons to press? Is there someone who always seems to know what you're thinking, or mirrors back your own fears and desires in a way that strangely seems to suit them?

Narcissistic Personality Disorder (NPD) is what psychologists call a trait of habitual lying and manipulation from otherwise intelligent people who seem to lack a part of their mind that cares about others. Someone lacking empathy or acting very selfishlu is sometimes labelled a "psychopath" or "sociopath".

The narcissistic psychopath (NP) watches very carefully and learns from things you do and say, with one goal, to manipulate and use you.

How do information threats work?

Didn't we just describe something familiar? They're called politicians!

Okay, not all politicians are that bad!

Joking aside, indeed faith in democracy and a less cynical belief in political life is a first line of defence against a very real enemy.

  • Demoralisation.

In psychological warfare, demoralisation is where an enemy sows seeds of hopelessness and despair amongst the population. Elections are a sensitive time to do that. If you think about it, getting everybody to stay at home and stop caring is a magnificent win for anyone who is against democracy. Sowing "learned helplessness" around issues of policy, technology, control and creating narratives of an "inevitable dystopian and authoritarian future" has been an ongoing strategy.

There's another reason to raise doubts about the conduct or outcome of an election:

  • Agitation.

Winding everybody up and stirring the pot is a good way to sap a nation's strength. The goal is division, to get friends and neighbours fighting. This works better if people are already cut-off from each other. Which leads us to:

  • Isolation

The aim is to make people feel disconnected from their fellows. from their work, common interests, and from their community and country. The fancy word is "alienated". Digital technology, housing poverty and low mobility already achieves this splendidly. Isolation is corrosive to democracy.

  • Dehumanisation

Attacking people's capacity for empathy and interpersonal relations is a "divide and conquer" strategy. Creating monsters, outsiders, labelling of everything as "extremist" and fomenting fear is characteristic of this sort of disinformation.

  • Iconoclasm

An "icon" is a respected symbol. Iconoclasm is the destruction or defaming of important symbols. By undermining the legitimacy of institutions, particularly civil institutions that offer checks and balances against the aims of disinformation agents, the public sense of cohesion and "anchoring" is disrupted.

But are they real?

Back to the strange similarities among the worlds of Narcissistic Personality Disorder, political life, and AI.

A real politician is a human. He or she comes to win you over by visiting your house or at least writing pamphlets. You can meet them face to face, and verify they are real. They'll even shake your hand to help you remember that. You can see that the promises in their pamphlets are the same as the ones delivered to your neighbours.

That's traditional politics.

In this century political tactics changed. In the past, broad policies were presented to everyone on television. That's changed to a practice of "micro-targeting", which smartphones and social media make easier.

Like advertisers, politicians started saying whatever they thought you would like to hear privately, whilst telling quite a different story to your neighbour.

Since we are isolated and talk less outside our "social bubbles", they don't get caught out. And if they do, they can just lie about it, delete Whatsapp messages, make policy documents disappear… because as we see more and more - technology facilitates fraud and corruption. That's because it is essentially opaque to ordinary people. It can give the appearance of almost anything.

We even have very outdated laws in the UK that presume the correctness of digital computation - completely counter to the advice of the best software engineers and computer scientists who know that computers almost always have bugs.

That's why we don't and should not trust electronic voting systems in the UK.

So imagine if someone turned up at your door, having just read a dossier of your personal concerns, likes, hates and financial history. Would they be at an unfair advantage to promise you things that would get your vote?

But they don't know me

Actually they know you quite well. If you've ever searched online, or bought something with a bank card you're at risk. So that's… well everybody.

Public relations and influence companies let political movers buy data and services to manipulate you. Although the story of Cambridge Analytica caused a "scandal", nothing really changed, and these kinds of businesses are more common than ever. Their methods and technologies have advanced and spread to other areas.

Commercial surveillance has been a thing for over two decades now, with companies amassing your data, trading it on a shady market where giant advertising and analytical firms operate. This data is powerful in the hands of targeted manipulators who know exactly how to get under your skin.

Influence is a multi-billion pound business. Advertising is the friendly face of much bigger industrial and military operations all around the world. Despite laws to regulate them, companies and foreign governments still brazenly lie about data they collect and how they use it.

Of course most of us don’t trust corporations, and have no reason to. It is not at all cynical that we expect them to act ruthlessly, to put profit above all else. They just want to sell us stuff and get rich.

We expect more integrity from public servants and institutions. Indeed a vital function of government used to be regulating industry.

But government regulation of private surveillance and mass influence fully failed. In part, that's because of surveillance alliances between big-tech and government. In part it's also because we love free speech. It's because we're liberal. It's to our credit that we allow our worst enemies to speak their mind, confident that we can safely dispute them. It's to our credit that we can live quite openly, including feeling no shame for our boundaries and rightful desires for privacy. To anti-democratic forces, our tolerance and openness are our greatest weaknesses.

So what about AI?

Imagine an army of obsessive narcissistic psychopaths, one for every person in the country. Your own personal lunatic stalker.

Until now there haven't been enough campaigners in any party to actively use detailed personal information and analysis effectively. That all changed with AI.

Current AI is technology capable of generating words, pictures or sounds that seem to mimic a real person or events. It can create realistic films and photographs of things that never were, or tell stories to suit an agenda.

More importantly, it can interact with you, responding to questions or concerns in a way that seems no different than a real person. AI is fake people. But unlike real people who are slow to recruit and motivate in large numbers, AI influencers can be created by the millions in datacenters around the globe. They can:

  • make or respond to social media posts
  • listen to conversations and follow threads
  • gauge your sentiment and feelings
  • predict your actions, responses and moods

Right now there are thousands of machines out there that:

  • contain some understanding of you as a person
  • are determined to reach you, and have your contact details
  • are set on changing your views
  • are programmed to leverage your emotions
  • are programmed to be deceptive and manipulative

They are electronic NP's. Narcissistic psychopaths in a box.

Advertisers and cybercriminals have long abused this technology, either to get you to buy things you don't want or need, or to trick you into installing malware through "phishing" attacks.

Now they are in the hands of groups, foreign and domestic, legitimate, criminal or hostile, who will use them during the next general election to affect British politics to their own ends.

Weaknesses of AI enemies

But there are things they cannot do:

  • care about you as a person, or anything else
  • have any morals, particularly about lying
  • hold any actually intelligent vision, opinion, idea or policy
  • have any accountability
  • physically inhabit a human body

If nothing else, this maybe helps us to spot them.

All the same, they can fake or lie about any of the above in order to gain your confidence.

Imagine a predator that will do anything it can to get you to behave as it wishes, by a mix of spying and lying. It knows your weaknesses and is programmed to exploit them. The best defence you have is to spot its approach and avoid it.

Shouldn't such technology be illegal?

As with all cyberweapons, this kind of AI is hard to regulate because the technology has good uses too. Computer scientists say that spies, criminals and advertisers are giving AI a bad name, which may hurt its future use in medicine, education and governance.

Despite new laws being drafted, presently there's little to stop anyone from using AI-based automated and targeted mass-manipulation technology.

Money talks

AI amplifies the effect of money on politics. Just as the internet changed surveillance, AI changes manipulative influence. It makes far reaching influence available to anyone with the financial and technical resources to run AI campaigns.

So AI is not a "leveller" in this sense. You or I might use a laptop to photoshop a fake image for a funny meme and post it online, but we cannot afford the resources to run a datacenter full of algorithms targeting millions of individuals around the clock. AI never sleeps. This is the problem of scale and capital advantage.

With AI getting easier to set up, train and quickly deploy, a greater threat therefore comes from already big players like hostile nations, terrorist groups and large industrial corporations. As with lobbying, campaign-spending favours those who already have disproportionate wealth and networks of connections. Or simply those who can buy capabilities.

Smaller circles

Your "circle of trust" as Robert De Niro's funny character in Meet The Fockers puts it, is something you need to keep an eye on.

  • Consider how well you really know some people, especially if they are very vocal at this time.
  • Value information more from people you've known a long time.
  • Weight information according to its diversity of independent sources.
  • Make more effort to connect with people you do trust and love, but you disagree with their politics. They may offer valuable perspective.
  • Geographical distance isn't a factor. Indeed it's good to get outside perspectives from people abroad that you know well and trust.
  • Individuals from apparently hostile nations are rarely implicated, so don't allow healthy distrust of information to become racism. Your Russian, Chinese, Korean and Islamic or Israeli friends, if verifiably human, remain human and remain friends due respect.
  • Organisations are not the same as people. They have no mind or "soul", and will treacherously chase their own needs first. Much as you may love your favourite brands like Apple, Google or Facebook, they are not reliable and trustworthy sources of opinion.

In general, try to cultivate a healthy scepticism about everything you see, hear or read from digital sources and value face to face conversation with real, diverse people foremost.

Trust but verify

If you've ever contacted your Member of Parliament, if you forget to tell them your name and address they usually ask you for that. If you've ever tried to do work in the UK, or contacted your doctor's surgery to query an appointment, more and more people demand proof of who you are.

That's actually a very difficult thing to do in this time. And digital technology makes it worse not better. People aren't really sure what constitutes "identity" or "access permission". ID documents don't really help, are cumbersome and become a "spoofable" target that people place too much trust in.

Have you ever been asked to send a high resolution scan of your passport through an unencrypted email like Google? Welcome to the club of billions of people exposed to higher risks of identity fraud.

Britain is becoming a "low trust" society.

In order to re-establish trustworthiness, we have to demand the same level of scrutiny and evidence from politicians, media outlets, services and organisations.

Knowing how to check website certificates, the WHOIS database and verify using Companies House or public HMRC records is a useful skill. But it's only a first line of defence against disinformation.

Calling someone back on a known phone number is often a good precaution. And people can be looked up on public electoral registers to check against a proof of address.

All public offices process old fashioned mail. They don't want to. It's "time consuming and inefficient". But they're legally obliged to. So writing letters is a superpower when it comes to unmasking and questioning anything that hides behind a veil of digital mystery.

But what about online?

In some cases it's almost impossible to find who is behind an idea, or news story. If you cannot trace information to any identifiable and respectable source, an individual or group with visible "skin in the game", then treat it as no more than a rumour and probably false.

Media Moratoriums

An idea practised in some countries is having a break in the weeks before any election, where no media related to controversial domestic politics is allowed. In the age of social media this is not possible, since no nations can really control the feeds from Facebook and the like.

What we recommend is self-chosen abstinence from the media firehose to give yourself reflective space to think things through alone.

  • Switch off all notifications and "news" alerts.
  • Visit your favourite discussion groups less - take a break for a week.
  • Share more cautiously. Forwarding memes and funnies is part of life, but just before an election you may unwittingly contribute to the spread of misinformation.
  • Be especially cautious of visiting new internet sites in this period.
  • Pause before following sensational "clickbaity" headlines.

Am I at risk?

You are more at risk if you are a "swing voter", meaning that you are still undecided about your vote.

You are more at risk as an "influencer", prominent person or in a position of responsibility. Be a little more cautious about repeating unverified political gossip to your co-workers, employees and associates.

You are more at risk if you have a long history of online social activity, use supermarket loyalty cards, rarely use cash, and generally if you have a quite trusting relation to technology gathering your data.

What is the goal of messing with our politics?

It is tempting to imagine specific groups trying to push their agendas. Indeed, this is part and parcel of ordinary politics. There is no particular political party, or cabal of moustache-twisting industrialists who, alone, pose a supreme danger.

The danger comes from undermining belief in democracy itself, to make people doubt the legitimate outcome of elections, and to lose faith in the idea of ground-truth. Everything from economic weakness to civil instability are the desired results.

This is a destabilising, long-term or strategic form of warfare. It is also ideological, and rooted in authoritarian and anti-democratic ideas that are gaining strength in the world today.

Tom Tugendhat MP acknowledges the irony (yet non-contradiction) that the fruits of liberal democracy, whether free speech or corporate power, are themselves part of the problem. Less obvious is the way that governments themselves contribute to the demise of democracy when warrantless mass-surveillance, ostensibly in the name of crime-fighting counter-terrorism, is normalised.

We also have to look closely at our own latent authoritarianism if we are to fight it on the world stage.

Links

Read more about the National Cyber Security Centre and "defending democracy" here:

https://www.ncsc.gov.uk/collection/annual-review-2023/resilience/case-study-defending-democracy

A good speech by Tom Tugendhat MP. Maybe starts off a bit flag wavey but once the sound of trumpets fades what remains is the clarity with which he nails the general threat. As security minister his POV prioritises state actors but he does acknowledge commercial, international criminal and terrorist threats as disinformation sources.

https://www.gov.uk/government/speeches/defending-democracy-in-an-era-of-state-threats

All democratic countries are affected. Read more about the thinking in Europe here.

https://v-dem.net/our-work/research-programs/democratic-resilience/

https://www.europarl.europa.eu/RegData/etudes/STUD/2021/653635/EXPO_STU(2021%5C653635_EN.pdf

And for a stateside view of things:

https://www.ncsl.org/elections-and-campaigns/artificial-intelligence-ai-in-elections-and-campaigns

Date: 20 January 2024

Author: H. Plews, E. Nevard, A. Farnell

Created: 2024-02-01 Thu 12:21