Do you want to use AI or be used via AI (2/2)

In my previous post (1), I showed how state actors who have been using propaganda for years to influence what we as individuals think/believe are now getting more aggressive, more targeted and more insidious by using AI/ML bots and social media.

It is still my belief that:

• Influencing how people behave is nothing new, AI just makes it easier to achieve
• AI is being used more and more to influence how people think and therefore behave
• AI is neither good nor bad, it is how it is used that matters.
• Each one of us has a choice to make, whether to allow the influence or question it

AI is neither good nor bad, neither was fire. It is how it is used that matters, and how we, individually deal with it. For example, it probably is not usual for you to leave flammable stuff next to a gas stove/BBQ, is it? You want to control fire and use it, rather than let is go loose because of error/clumsiness of mind, and cause untold damage.

The battle is for your brain, do you want to lose it?

perchance.org

The most important defense is a sane level of skepticism

People are wired to trust (1). When people’s social circles were small, and tight, you knew the people you interacted well, and their personal characteristics. Add to this mutual benefit (after all you are in a tight knit group), it made sense to trust information that someone in the group relayed to you.

However, the world has evolved from small villages, to global village, add to this the evolution of the press from reporting to giving opinions, and the advent of social media where pretty much anybody can say anything, and you can see that it makes sense to be a little bot skeptical of information you get.

It is ok to be skeptical, but just parking information somewhere is not a good idea.

• It may be believed since it has stayed in the parking lot for that long or with similar stories
• You may be suspending action while you are evaluating things (analysis paralysis)

Aside: how processing information has changed over time

This is an aside that is slightly more important than a footnote. One of the interesting things that has changed is HOW people are convinced. In the past, people used to postulate “a theory”, then provide ‘facts’ that support the theory, and therefore the people targeted believe the theory. Nowadays, the focus is in repetition rather than a theory. “There is a low bar to assent and repetition: the conspiracy charge is ‘true enough.’ Subscribers to conspiracy without the theory do not to have to believe the particulars of the charge. Indeed, there are rarely any particulars – just assertion and innuendo.”(2)

Nowadays repetition in social media is designed to burn a factoid into your brain, rather than bring you to a conclusion by starting with a theory. The end result is that a belief or a bias is formed, and given we are lazy, we rarely review the reasons for beliefs we may have. So, repetition, especially at the scale of social media, works.

Back to the main topic.

Simple Steps to protect yourself: 1 trusted sources

I started by using US elections of 2016 as an example (3), and this was deliberate. These two posts were motivated by an article I read recently. (4). The Stanford Internet Observatory was one of the first to pull the alarm on Russian interference in the US 2016 Presidential elections. If you have time, it is worth looking into their efforts (5). Basically, they are stuck in lawsuits, so the university has decided the truth was not worth that amount of money.

There are other possible trusted sources that evaluate factoids you may have heard, fact checking websites; politifact is one of them (6); you can find more (7)

Unfortunately for us outside the USA, many of these are US centric.

For example, how many people in Singapore would have believed, despite all the technological excellence of MAS, the banking sector (including the world’s best digital bank), the rapidity of the police at shutting down crime, SGD3Billion worth of money laundering took place (8). But it is a fact.

 

Simple Steps to protect yourself: 2 learn logical thinking

Someone once said common sense is less common than you think. I unfortunately agree. We were all given brains and these are our quite well equipped to defend themselves against ideas/concepts that are not correct. And logic is the tool that helps. The most important thing is of course to start by being skeptical, and take the time to think through things logically.

The first thing you should do is strip any statement to its basic components, like deconstruct something made of Lego, break it down to its basic components. You take the argument that has been proposed to you into its basic premises. Then you apply a logic technique to see where that gets you.

Deduction (9)

Deduction is the most reliable way to prove an argument or reject it. The key to deduction is to place the premises of an argument into a certain shape, and if the shape holds, then the argument is valid. For example: 

• All B are A
• C is B
• Therefore, C is A

Now let me flesh this as an argument with premised and conclusion:

• All humans are mortal
• John is human
• Therefore, John is mortal

How about this one?

• All women have long hair
• Jane is a woman
• Therefore, Jane has long hair

What do you think of this one? The form is the same as above, but do you agree? You probably wouldn’t. The argument is valid, however, one of the premises is not True.

• Not all women have long hair

While the shape of the argument is valid, because one of the premises is not true, therefore the argument itself is not sound; you cannot say with certainty that Jane has long hair because one of the premises is not true (Some women do not have long hair).

And that’s the thing about deduction, it provides as close to certainty as you can get, it is bullet proof.

You can see deduction as:

• Take an accepted Theory/Proven Idea
• Apply it to a specific case that fits
• Generate a conclusion for that specific case

What if you have no accepted theory/proven idea?

Induction (10)

While deduction is bullet proof, induction is less so; it gives you a probabilistic answer.

is really yummy, no tummy issues

Let’s say that, like me, you are a fan of la-zi-ji (spicy chicken).

• I ate la-zi-ji 10 times in the last 3 months and each time, I had tummy issues the next day
• The spices in La-zi-ji can be potent to cause tummy issues
• Therefore, causes tummy issues

 There are a few differences compared to deduction.

Here you start with some observations (eat la-zi-ji day 1, tummy issues day 2, 10 times in last 3 months), apply some analysis/theory (la-zi-ji spices may cause tummy issues) and derive a probabilistic conclusion.

What do you think of this induction, how strong is it?

To me it is not that strong. It can be improved by gathering more data:

• Eat from different stalls to make sure it is not stall specific
• How did your friends who also at la-zi-ji react? Or is it just you?
• Was It really la-zi-ji, what else did you eat, maybe you also ordered cray fish and you have a mild allergy

In induction, you start from observations, samples and try to build a theory. How strong the theory depends on the size and quality of your observations/samples.

The other important thing here is, if the 11^th time you at la-zi-ji, your tummy is ok the next day, then you can revise your theory to la-zi-ji may cause tummy issues, you lower the probability.

Abduction (11)(12)

No, I am not referring to kidnapping. Abductive reasoning is another logical tool you should equip yourself with.

Interestingly, this is the type of reasoning AI finds less easy.

Abduction starts with an observation, then you think of the various things that could explain the hypothesis, and postulate the one you find most likely.

An example will clarify things:

• You find crumbs on the kitchen table and unwashed plate and glass in the sink
• You think it is most likely that your flat mate had a midnight snack and was too lazy to clean up.

There are may other explanations that could fit:

• You sleepwalk and had a midnight snack while asleep last night
• A burglar came in and decided to have a midnight snack
• Your flat mate is playing a practical joke on you

You simply picked the one you thought most likely, based on your experience.

It is important to be able to recognize which of these paths of logical reasoning you have taken, then you can understand how much certainty you can associate with your beliefs based on which process you used to reach them.

It may be confusing at first, but with some practice you will get it.

Furthermore, there is a clear progression from Abduction to Induction to Deduction

This is aligned with how discovery takes place and models are created and built all the way to implementation.

The extra problem in the fight for the real estate of your brain is that some people/bots are really good at masking bad arguments, tricking you in believing they are correct.

 

Simple Steps to Protect yourself 3: Identify Fallacies

I always say I am lazy; I actually think humans are lazy and it takes conscious effort sometimes not to be lazy. Fallacies are shortcuts that people may choose to take advantage of your laziness and implant some stuff surreptitiously into your brain.

There are many fallacies, I will only illustrate a couple of the most commonly used:

Appeal to Authority / Ad Verecundiam

This simply uses a famous person as authority figure to justify the argument. There are 2 flavours: False Authority

• CR7 uses K water
• CR7 is world famous
• Therefore, K Water must be a great soft drink.

Nobody said CR7 was an authority in water, this is the world of influencers…

Anonymous authority

• Scientists claim 5G is harmful
• Scientists know what they are talking about
• Hence 5G is harmful

Who is the scientist, where are the scientific papers?

Bandwagon/Ad Populum

People who have known me well are likely to have heard me respond angrily to Ad Populum arguments; they are basically: do this because everyone does so… No need to think, just follow the bandwagon.

• Everyone buys extended warranties
• Extended warranties are worth it
• Therefore, I should buy one on my spice grinder

 Strawman

A strawman argument is used when someone’s argument is distorted by an opponent so that when the distorted argument is refuted by the opponent, they claim the original argument has been refuted. It can take many forms: over simplifying what someone said, taking it out of context… It is used very very often by politicians

Example:

• I am against the mandatory sentence for drug mules
• He is for allowing entry of drugs into the country

False Dichotomy

False dichotomy truly plays on laziness. It aims to wrongly simplify an argument into 2 possible solutions, one of which is not favourable so you pick the other, ignoring all other solutions.

Example:

• Either you eat an apple a day, or you visit the doctor.
• You choose.

Ad Hominem

Ad Hominem occurs when the person is attacked rather than the argument that the person is making.

Example: 

• I don’t think this person can be trusted because he made a mistake last time

This has no impact on the argument the person presents, but the ad hominem attack to meant to damage the credibility of the argument by damaging that of the person making it.

There are many more fallacies, I have just highlighted a few I am sure that, if you think about it, you have seen in action before. You can find more details; Wikipedia is a good place to start (13).

Conclusion

The best way to minimize chances of your brain being taken over by ideas/concepts that are not that correct is to learn to identify these ideas/concepts/arguments/snippets of ‘information’ as early as possible. You can rely on public service websites that check for truth, or choose to verify the veracity/soundness/validity of arguments by your own self, or both.

It is equally important to verify your beliefs in light of new data you may come across and update your beliefs/ideas. It is a constant battle, probably worth being engaged in.

 

1. https://www.linkedin.com/feed/update/urn:li:activity:7216242124814524417/
2. https://www.psychologytoday.com/us/basics/trust
3. https://www.researchgate.net/publication/337877932_Conspiracism_and_Delegitimation_A_lot_of_people_are_saying_The_new_conspiracism_and_the_assault_on_democracy_Russell_Muirhead_and_Nancy_Rosenblum_Princeton_University_Press_Princeton_2019_978069118883
4. https://www.theguardian.com/commentisfree/article/2024/jun/29/closing-the-stanford-internet-observatory-will-edge-the-us-towards-the-end-of-democracy
5. https://cyber.fsi.stanford.edu/io
6. https://www.politifact.com/
7. https://library.csi.cuny.edu/c.php?g=619342&p=4310783
8. https://www.straitstimes.com/singapore/courts-crime/3b-money-laundering-case-su-jianfeng-sentenced-to-17-months-last-of-10-to-be-sent-to-jail
9. https://en.wikipedia.org/wiki/Deductive_reasoning
10. https://en.wikipedia.org/wiki/Inductive_reasoning
11. I am using the classical definition of abduction here, as first explained by Pierce https://plato.stanford.edu/entries/abduction/peirce.html
12. For a modern definition, you can consult https://plato.stanford.edu/entries/abduction/  ; however, I think the classical definition works best in the process of formulating ideas and also as part of the process of Analytics.
13. https://en.wikipedia.org/wiki/List_of_fallacies