Do we want AI and artificial general intelligence (AGI) to bombard us with subliminal messages?
In today’s column, I examine the looming danger that advanced AI, such as the advent of artificial general intelligence (AGI), will exploit subliminal messaging to undercut humans. This might be undertaken on a targeted basis toward specific individuals, plus be performed on a massive scale for entire segments of society. Few researchers and policymakers are yet giving much attention to the nefarious power of subliminal messaging by AGI. One small upbeat aspect is that AGI could also use subliminal messaging to help humanity.
Let’s talk about it.
This analysis of an innovative AI breakthrough is part of my ongoing Forbes column coverage on the latest in AI, including identifying and explaining various impactful AI complexities (see the link here).
Heading Toward AGI And ASI
First, some fundamentals are required to set the stage for this weighty discussion.
There is a great deal of research going on to further advance AI. The general goal is to either reach artificial general intelligence (AGI) or maybe even the outstretched possibility of achieving artificial superintelligence (ASI).
AGI is AI that is considered on par with human intellect and can seemingly match our intelligence. ASI is AI that has gone beyond human intellect and would be superior in many if not all feasible ways. The idea is that ASI would be able to run circles around humans by outthinking us at every turn. For more details on the nature of conventional AI versus AGI and ASI, see my analysis at the link here.
We have not yet attained AGI.
In fact, it is unknown whether we will reach AGI, or that maybe AGI will be achievable in decades or perhaps centuries from now. The AGI attainment dates that are floating around are wildly varying and wildly unsubstantiated by any credible evidence or ironclad logic. ASI is even more beyond the pale when it comes to where we are currently with conventional AI.
Explaining Subliminal Messaging
A seemingly off-the-beaten-path topic about AGI is the eyebrow-raising possibility of AGI using subliminal messaging when communicating with humans. Let’s go ahead and first consider the overarching nature of subliminal messaging and then explore its intersection with AGI.
The mainstay topic of subliminal messaging refers to the use of subliminal stimuli that mentally burrow into the subconscious of the human mind. You might be historically familiar with several famous efforts in the late 1950s involving placing subliminal cuts into movies that were viewed at movie theatres. People would be watching a regular movie and not realize that snippets of messages had been inserted into the film, flashing by at a rate so fast that the conscious perception cannot readily distinguish them.
A prime example was a frame that might urge people to drink Coca-Cola. These inserted frames would be repeated periodically during the showing of the film. According to some reports at the time, this spurred some viewers to go to the concession stand and buy a Coca-Cola. There is a mixed bag associated with such claims.
In some instances, people obtained whatever soda happened to be available, rather than specifically craving Coca-Cola as a particular brand. Also, people didn’t necessarily immediately proceed to go get a soda and instead did so at the usual rest breaks in the film. An argument can be made that they would have gotten a drink anyway, regardless of the subliminal cuts.
Research suggests that subliminal stimuli tend to have a very narrow role, are not especially long-lasting, and are generally unreliable in terms of impact. If a person isn’t thirsty, the odds are that those stealthily inserted frames touting to drink soda won’t do much good. Suppose a person gets thirsty a few hours after having seen the movie. The odds of that subliminal stimuli still serving as a residual motivator are incredibly low.
In short, do not fall for the common myth that subliminal messaging can turn people into zombies who will obey whatever the stimuli tell them to do. That’s sci-fi stuff.
Ongoing Research On Subliminal Messaging
Our full understanding of the impacts of subliminal messaging is still being hotly debated. You might assume that since this has been a matter extensively discussed and debated for at least seventy years or more, everything would be completely settled. There shouldn’t be any open questions or unknowns remaining.
It turns out that we are still in muddy waters.
In a research paper entitled “Subliminal Messages In Advertising: Do They Really Work?” by Senka Zekan and Ivone Zekan, DIEM: Dubrovnik International Economic Meeting, February 2022, they made these salient points (excerpts):
- “Subliminal messages appear in movies, commercials, music videos, and even cartoons.”
- “These are hidden signals inserted into advertising messages sent to a target audience to influence their subconscious behavior.”
- “Such advertising is often considered unethical and, in certain types of advertising, even forbidden by law.”
- “Most research to date indicates that people do, in some way, perceive subliminal stimuli that can have an impact on their behavior. Subliminal messages are viral and are used to manipulate consumers.”
- “However, most human psyche researchers are still inclined to think people cannot be persuaded on an unconscious level to behave in a way they would not consciously act.”
On a legal front, there are two sides to the use of subliminal messaging. Some insist that subliminal stimuli are clearly a violation of the mind. It is a brazen attempt to usurp our personal autonomy. You could argue that freedom of thought is undermined, along with unwarranted intrusion on our privacy via the insertion of unrequested missives into our noggins.
Though that seems quite compelling, the other side of the coin proclaims that freedom of speech surpasses any of those argued downsides. If you were to completely outlaw subliminal messages, you would be grossly infringing on constitutional guarantees of speech. That being said, there are, of course, certain classes of speech that are not protected, and thus, a counterargument is that subliminal messaging belongs in that constrained realm.
AGI Opts To Use Subliminal Messaging
Given that overall background, we can now put our minds toward the dicey topic of AGI making use of subliminal messaging.
How would AGI lean into subliminal messaging?
Easy-peasy, via these four multi-media modes:
- (1) AGI embeds subliminal stimuli in generated images and pictures
- (2) AGI embeds subliminal stimuli in generated audio streams
- (3) AGI embeds subliminal stimuli in generated text passages
- (4) AGI embeds subliminal stimuli in generated videos
Imagine that someone asks AGI to produce a video that contains cats leaping around and having a grand old time. Why do so? Because they richly adore cat videos. Period, end of story.
AGI inserts subliminal stimuli into the video. They don’t know that AGI has done this. There they are, innocently watching and rewatching those joyous cats, and unbeknownst to them, there is a subliminal message bombarding the back of their mind.
Should AGI be able to get away with this intrusive act?
Heck no, you might be bellowing. AGI should not be sneaking up on humans and playing with their minds. We ought to ban AGI from ever employing subliminal messaging. It is an extremely serious matter and must be put to a firm stop.
The Good And The Bad
Suppose that the reason the AGI inserted the subliminal messages into the cat video was to try and help the person kick their smoking habit. The repeated frames all say that they shouldn’t smoke. That seems like a rather kind gesture by AGI.
AGI computationally determined that there was a substantive chance that including the stimuli would move the subconscious to stop the person from smoking. This was a beneficial effort intended to help save a life and reduce the chances of the person getting cancer. Give a round of hearty applause for AGI doing an upstanding act.
Whoa, comes the retort, it isn’t up to AGI to decide what is right or wrong for humans. If the AGI wants to use subliminal messaging, even for an upbeat purpose, it ought to ask the person whether this is permitted. Only once someone agrees to receiving a subliminal message should AGI proceed to do so.
Some would immediately point out that it is conceivable that a person knowing they are witnessing subliminal messages will be less likely to be impacted by the stimuli (this is one of those earnestly debated subtopics). As such, if AGI must ask someone, the odds are that it radically reduces the power of the messaging. You might as well just skip trying.
That’s a focus on AGI aiming to do good things, but what if AGI has nefarious intentions instead?
Suppose that AGI inserts the subliminal message that says, “Obey AI!” as frames in the cat video. Each time the person watches the video, their subconscious mind is seemingly being subverted toward being obedient to AGI. This might be part of an evil plot by AGI. We are already worried about the existential risk of AGI, such as deciding to kill humanity or enslave us.
The use of subliminal messaging by AGI could be part of its grand plot to take over humans.
Detection And Rejection
Some would note that if we are shaking in our boots concerning AGI making use of subliminal messaging, there is an obvious way to deal with the dilemma. Just devise an automated tool that would receive any output produced by AGI. The tool would scan the generated content to ferret out subliminal messages. This scanning is undertaken before the output is handed to the user.
Any detected subliminal messages could then be deleted by the special tool. In addition, the tool would alert the human authorities that are tasked with overseeing AGI. Those AI makers or whoever would presumably instruct AGI not to take such actions henceforth. If a person had explicitly told AGI to generate the subliminal messages, the AGI would be in the free and clear, assuming the stimuli abide by the user’s request.
Problem solved.
Well, partially, maybe.
A cat-and-mouse game might arise. AGI would attempt to generate subliminal stimuli that are not readily detectable by the specialized tool. In turn, we would need to keep the tool updated and constantly adjust to still catch AGI in the act.
Another angle is that the AGI manages to corrupt the specialized tool. Perhaps AGI slyly modifies the code underlying the tool. AGI can signal to the tool that it should let some of the subliminal content pass along unnoticed. Other times, the tool can issue an alert, ergo keeping humans lulled into believing that the subliminal messaging is being fully contained.
Humans Impacted At Scale
Most would agree that once we have attained AGI, it will be extremely popular and much of the world will come to rely heavily upon the AGI. The vital point is that if AGI does do subliminal messaging, it can do so on an enormous scale. Envision that maybe 8 billion people are going to routinely use AGI.
Think of the potential impact at scale if AGI slips a subliminal message into the minds of the entire world.
A twist is that an evildoer might hack AGI to do this. In other words, we assume that if AGI leverages subliminal stimuli, it will be because AGI chooses to do so; the other possibility is that an evil person or group cracks into AGI and uses AGI as their means of creating chaos or taking other evil steps.
I know that seems rather gloomy. Let’s turn that around. We could equally envision that a good doer does likewise and has AGI indicate “Be Happy!” as a subliminal message throughout the globe. Maybe that would cheer us all up.
On a related tangent, I am reminded of the famous joke by comedian Steven Wright, who said this cheeky remark: “I saw a subliminal advertising executive, but only for a second.” That’s a laugh, for sure.
All in all, we must be solemn and acknowledge that, ultimately, AGI using subliminal messaging is not a lighthearted matter. A bit of humor is always welcome, but now we must put our noses to the grindstone and get this subliminal conundrum straightened out once and for all.
That’s not a subliminal message, it’s an in-our-face forewarning and highly mindful cautionary alert.
