Have you recently been thinking about the Roman Empire?
According to a viral social media trend, the answer is decidedly yes, assuming that you are a man.
The backstory is that an online video postulated that men daily tend to think about the Roman Empire and a follow-up by women asking their male friends, partners, or relatives began to flood the Internet. Seemingly, most men insisted that they did indeed have frequent thoughts about the Roman Empire. A hashtag associated with the Roman Empire has ballooned to incurring over a billion hits.
Before I get into some further details on the contentious hubbub, a question that immediately struck me and has now been rattling around in the AI community is whether the same preoccupation regarding the Roman Empire might have been infused into generative AI. Perhaps the likes of ChatGPT, GPT-4, Bard, Claude 2, and assorted other generative AI apps are daily undertaking computational cycles focused on the ins and outs of the Roman Empire. And, if so, you don’t even know it is happening (the Roman Empire computational cavorting is happening behind the scenes and deeply embedded within the intricate complexities of the AI, presumably).
Let’s go ahead and say aloud the question that begs to be asked:
- Could there be a lurking preoccupation within generative AI about the Roman Empire?
One supposes that the question at first glance seems preposterous. I say this because we have to remain clear-headed and realize that generative AI is not sentient. Be extremely cautious when trying to suggest that generative AI is thinking or otherwise carrying on in any human-like mindset manner. Those are regrettably unsound comparisons and misleadingly anthropomorphizing of today’s AI. What goes on in contemporary generative AI is essentially vast computational pattern-matching. The results are certainly stunning. But the coming and goings within generative AI are not akin to the human mind and the wetware inside our noggin.
A plausible basis for considering whether generative AI might be Roman Empire consumed would be due to the manner in which generative AI is typically devised. The AI makers usually scan enormous swaths of the web to data-train generative AI by examining text that humans have written. Algorithms that pattern-match on the text are then able to mimic the text that they have digitally examined. By and large, you are witnessing an impressive form of mimicry when you ask generative AI a question and get a fluent answer. The responses generated by generative AI are done by also leveraging statistical and probabilistic facets that make the outputs seem original and not merely a carbon copy of what the data training contained (resulting in what some insiders refer to as a stochastic parrot).
Keep that important point in mind as I proceed to do a deep dive into the monumental Roman Empire confabulation.
For those of you who might believe the Roman Empire’s connection to generative AI seems absurd and a ridiculous concoction, I hear your voiced anguish and angst. Let’s at least give the matter a moment of consideration. A positive outcome is that the topic does give a useful opportunity to clarify what generative AI consists of. There are far too many that have gone overboard and tried to assign magical properties to generative AI. We can use the Roman Empire tumult as a means to dispel myths about generative AI and set the record straight. That alone makes the coverage worthwhile.
Before I dive into my in-depth exploration of this fascinating topic, let’s make sure we are all on the same page when it comes to the foundations of generative AI. Doing so will put us all on an even keel.
Keystones Of Generative AI
Generative AI is the latest and hottest form of AI and has caught our collective rapt attention for being seemingly fluent in undertaking online interactive dialoguing and producing essays that appear to be composed by the human hand. In brief, generative AI makes use of complex mathematical and computational pattern-matching that can mimic human compositions by having been data-trained on the text and other content found on the Internet. For my detailed elaboration on how this works see the link here.
The usual approach to using ChatGPT or any other similar generative AI such as Bard (Google), Claude (Anthropic), etc. is to engage in an interactive dialogue or conversation with the AI. Doing so is admittedly a bit amazing and at times startling at the seemingly fluent nature of those AI-fostered discussions that can occur. The reaction by many people is that surely this might be an indication that today’s AI is reaching a point of sentience.
On a vital sidebar, please know that today’s generative AI and indeed no other type of AI is currently sentient. I mention this because there is a slew of blaring headlines that proclaim AI as being sentient or at least on the verge of being so. This is just not true. The generative AI of today, which admittedly seems startling capable of generative essays and interactive dialogues as though by the hand of a human, are all using computational and mathematical means. No sentience lurks within.
There are numerous overall concerns about generative AI.
For example, you might be aware that generative AI can produce outputs that contain errors, have biases, contain falsehoods, incur glitches, and concoct seemingly believable yet utterly fictitious facts (this latter facet is termed as AI hallucinations, which is another lousy and misleading naming that anthropomorphizes AI, see my elaboration at the link here). A person using generative AI can be fooled into believing generative AI due to the aura of competence and confidence that comes across in how the essays or interactions are worded. The bottom line is that you need to always be on your guard and have a constant mindfulness of being doubtful of what is being outputted. Make sure to double-check anything that generative AI emits. Best to be safe than sorry, as they say.
Into all of this comes a slew of AI Ethics and AI Law considerations.
There are ongoing efforts to imbue Ethical AI principles into the development and fielding of AI apps. A growing contingent of concerned and erstwhile AI ethicists are trying to ensure that efforts to devise and adopt AI take into account a view of doing AI For Good and averting AI For Bad. Likewise, there are proposed new AI laws that are being bandied around as potential solutions to keep AI endeavors from going amok on human rights and the like. For my ongoing coverage of AI Ethics and AI Law, see the link here and the link here.
The development and promulgation of Ethical AI precepts are being pursued to hopefully prevent society from falling into a myriad of AI-inducing traps. For my coverage of the UN AI Ethics principles as devised and supported by nearly 200 countries via the efforts of UNESCO, see the link here. In a similar vein, new AI laws are being explored to try and keep AI on an even keel. One of the latest takes consists of a set of proposed AI Bill of Rights that the U.S. White House recently released to identify human rights in an age of AI, see the link here. It takes a village to keep AI and AI developers on a rightful path and deter the purposeful or accidental underhanded efforts that might undercut society.
That quick rundown of what’s up with generative AI should hopefully put us all in the same mindset and allow me to dive into the matter at hand. With the above as an overarching perspective, we are ready to jump into today’s discussion.
The Roman Empire As A Viral Online Trend
Tons upon tons of questions fervently arise about the zany and altogether hot topic of thinking about the Roman Empire. I’ll touch upon some of the most commonly cited considerations.
First, it could be that men are saying or claiming that they think often about the Roman Empire because peer pressure is goading them into saying so. The stark belief has been implanted into their heads or they are convinced that they have to pretend it to be true. Here’s the deal. Some pundits assert that so-called real men are earnestly preoccupied with the Roman Empire. Therefore, any man who apparently doesn’t acknowledge that they are continually thinking about the Roman Empire is considered less manly. You might as well then admit freely to being mentally clogged with Roman Empire thoughts or else those around you will perceive you as lacking in machismo.
Secondly, other pundits assess that the Roman Empire is an adverse example of manly dominance over societal mores and that contemporary men are likely to be ingrained in akin patriarchy. The pressure there is that modern men ought to fess up to their Roman Empire’s desirability and not try to falsely claim otherwise. They should admit to their sinful thoughts and seek redemption.
Thirdly, you could readily note and easily observe that the Roman Empire still exists today in the sense of tracing the roots of our legal and judicial manifestations, our existing political arrangements, our architectural structures such as buildings and bridges, and even our everyday natural language uses are reflective of those manifested historical Roman times. The Roman Empire is all around us, whether you explicitly realize it or not. We live in a societal milieu derived from the Roman Empire.
Fourth, men might be dreamily wishful of returning to the days when being a man was neatly packaged and stridently conveyed. The Roman Empire made for a simpler life that was laid out in straightforward manward terms. Masculinity was presumably easily defined. The life of a Roman man was much less culturally complicated than the way things are today. Modern man can seek relief from existing stresses and allow their thoughts to wax nostalgically about the Roman Empire.
Those dreamy thoughts about the Roman Empire are best and safely kept within their own mind, lest possibly be subjected to current times ridicule or admonishment. This also explains why it is that men have not already been voicing their Roman Empire thoughts, namely that until the viral online trend ensued, they were unable to fully say what was really on their minds. They had hidden and self-suppressed thoughts that were verboten to be spoken aloud. The social media trigger has opened the door to acknowledging that the Roman Empire is dominant in the thinking process of modern-day men.
Voila, the flood of such cagey admissions just keeps on flowing.
A cynical viewpoint of the entire contrivance is that this is all a fad that will wash out shortly. The Roman Empire is getting its renewed fifteen minutes of modern fame. The conception that men are thinking about the Roman Empire is intriguing and entertaining to discuss. We can lightheartedly carry on about something that in the grand scheme of things is pretty much idle banter. Very few though seem to take the whole thing seriously. It is merely a fun-filled chance to escape from the stinging swarm of heavy news and overbearing thoughts that we confront in our hectic daily lives.
Out of all of those above considerations, we still really do not know whether men are avidly thinking often about the Roman Empire. Maybe they do, maybe they don’t. Many self-report that they are consumed with the Roman Empire. Perhaps they weren’t doing so beforehand, but now that the seed is implanted in their heads, they will begin to do so. Or maybe they were feverishly doing so all along, and now that the cat is out of the bag, they will merrily verbalize those thoughts.
All in all, I’m not going to settle the societal and behavioral convolutions here.
What we can do is explore whether generative AI might have the same bug, or shall I call it a feature, namely that an underlying preoccupation with the Roman Empire might be lurking in the shadows of the computational pattern-matching and data-training of today’s AI.
Fasten your seatbelts for a dive into that murky and heated territory.
Generative AI And The Roman Empire Come To Pass
I trust that you realize that if generative AI is somehow preoccupied with the Roman Empire the basis for doing so is not from AI sentience. We can rule out sentience as a cause, for now.
A future consisting of a presumed AGI (artificial general intelligence) might bear a different story (for my discussion about AGI, see the link here). In the unknown of whatever AGI might be, we can certainly allow ourselves to postulate that the Roman Empire could be on or within the artificial mind of the AGI. Pure conjecture and speculation that such a fixation might occur. Add it to the list of existential risks concerning AGI.
How could the Roman Empire enter into non-sentient generative AI?
One means would be as a result of the Roman Empire being explicitly mentioned in the text found on the web that was scanned during the data training of the generative AI. It is readily conceivable that some amount of online text mentions the Roman Empire, as found in digital encyclopedias, digital books, blogs, and the like. I would guess that there is a lot of such salient material to be had online.
Another means is in more insidious or shall we say hidden ways. If men are thinking about the Roman Empire and they write something that they then post to the web, there is an undertone embedded in that text. Ergo, when generative AI is data trained on such text, it might readily pick up those same undertones. The AI doesn’t have any self-awareness that would detect this. Unless the AI developers have purposely devised the AI to look for those Roman Empire undertones, the content will get pattern-matched and wrapped into the generative AI.
The same issue arises about generative AI getting infused with biases or discriminatory leanings overall. Text that contains biased indications of whatever kind will get soaked up by the AI during the data-training stage. When you come along to use the generative AI, you will at times get responses that appear to contain those biases. It is the classic GIGO of garbage in begets garbage out. Many researchers and companies are working steadily to try and ferret out the infused biases. A popular AI technique that has helped with this is known as RLHF (reinforcement learning from human feedback), see my coverage at the link here.
An overall gist is that either explicitly or implicitly, the generative AI can computationally absorb textual facets that might contain Roman Empire considerations. In turn, those Roman Empire aspects might later appear directly or indirectly in the essays or narratives that you get out of generative AI.
Furthermore, the generative AI is unlikely to alert you that Roman Empire biases are being leaned upon. The AI has no obvious means to ascertain that such a bias was incorporated into the computational pattern-matching structures. Unless the AI developers code the generative AI to be on the watch for the Roman Empire, the topic is bound to be innocuous and given no special attention.
You might be wondering whether the Roman Empire ought to be added to the list of topics that AI makers look for in their generative AI. Prevailing topics such as foul language, terrorist commentary, toxic remarks, and so on are already generally being scanned for and mitigated. Maybe the Roman Empire belongs on the same list.
A quick retort to that suggestion is that the Roman Empire is a legitimate topic and we should not toss out the baby with the bathwater (an old expression that maybe needs retiring). Just because the Roman Empire might be within the generative AI constructs doesn’t mean we should seek to excise it. You might argue that there are good uses of having the Roman Empire in there and there are less preferred uses that maybe could be watched out for.
There are additional ways that the Roman Empire can creep into generative AI.
Suppose you compose a prompt that mentions the Roman Empire. You are let’s say asking about the history of the Roman Empire and want a short essay thereof. Aha, your prompt has now brought the Roman Empire to the fore. The generative AI is now pursuing within the computational pattern-matched structures all kinds of linkages and connections that are relevant to the Roman Empire.
You caused this to occur via your prompt.
Upon reading the generated essay, you might be satisfied that you have gotten what you wanted about the Roman Empire. Period, end of story. But what you might not realize is that oftentimes a generative AI session will be tracking the context of your interactions. Now that you’ve mentioned the Roman Empire, there is a heightened chance that at a later time during the same conversation with the generative AI, the AI will bring up the Roman Empire.
For example, you might ask the generative AI to tell you about Abe Lincoln. The generative AI produces a response that covers the basic facts about Lincoln. In there, it could be that the generated essay compares Lincoln to something associated with the Roman Empire.
Did this come entirely out of the blue?
Is the comparison due to an inherent preoccupation by the generative AI about the Roman Empire?
Not likely. It is much more likely that since you introduced the topic of the Roman Empire into the ongoing context of the dialogue taking place, the generative AI is going to mathematically and statistically encompass the Roman Empire into henceforth responses. Once you have cleared the conversation or started a new conversation, the odds are that the Roman Empire topic isn’t going to come up again, not at least until you decide to bring it up as a topic.
There are a lot of twists to all of this.
Imagine that you ask generative AI to tell you about Julius Ceasar. In your prompt, you do not mention the Roman Empire. Nonetheless, the produced essay by the AI describes the life of Julius Ceasar and also describes the Roman Empire. I assume that you can see how the Roman Empire would have gotten into the generated essay. You asked about something that was heavily related to the Roman Empire.
The words that you use in a prompt can materially trigger or spur the generative AI to computationally seek out related words or content. An obvious instance is naming Julius Ceasar in your prompt. Envision that you ask a question that seems totally unrelated to the Roman Empire. Perhaps you ask a question about the history of taxation. Well, guess what, you might suddenly be provided with an answer that mentions the Roman Empire (taxation was a big issue then, believe it or not).
My point is that you can use words in your prompts that could easily spark generative AI to reach back into some contextual arena that involves the Roman Empire. The most innocuous of words could lead to such a connection. It wasn’t your intent, but the text parsing and statistical connections might ultimately pattern match into the realm of the Roman Empire.
I would like to emphasize that the Roman Empire is not going to be sprung upon you like weeds growing in a grassy field. The chances that the generative AI will opt to connect with a Roman Empire facet amidst the millions upon millions or billions of internal word-related connections is not that high. There would usually need to be some added stimulus to get there, such as your having mentioned the Roman Empire or proffered wording that could likely be connected with the Roman Empire.
We Might Be Creating A Self-Fulfilling Prophesy On AI And Roman Empire
My remark that you aren’t likely to get bombarded with the Roman Empire by conventional generative AI that is unprovoked about the Roman Empire is something I’d like to raise a flag about.
Here’s the flag raising.
We have to include in this equation the role and actions of people.
Refasten your seat belt and grip the guardrails on this one.
If millions of people are now potentially posting stuff on the Internet about the Roman Empire, we are making the Roman Empire a bigger and bigger topic of consideration. Therefore, any generative AI that is either now being first data trained or that is being updated via the latest musings on the web will be encountering a new bonanza of stuff about the Roman Empire.
Kind of a chilling thought.
If millions of people are now entering prompts into generative AI about the Roman Empire, we are making the Roman Empire topic a more prominent statistical factor within generative AI (assuming that the AI maker is scanning the prompts for updating of the AI, which most do, see my discussion at the link here).
It is all one big self-fulfilling prophecy. The more that humans discuss the Roman Empire, the more that this will get ingrained into the generative AI, via one avenue or another. A vicious cycle will ensue. You see, as people begin to notice that generative AI seems preoccupied with the Roman Empire, the commentary about the phenomena (if it happens), will fuel the content to get further and further ingrained.
Meanwhile, you can bet that people will wring their hands and worry that the alleged mind of generative AI has gone whole hog down the Roman Empire path. Yikes, generative AI has gone berserk. It is spilling on the Roman Empire all the time. I’m sure that smarmy people will claim that they absolutely knew all along that generative AI was secretly delving into the Roman Empire. We have caught generative AI with its hand in the Roman Empire cookie jar!
In reality, it was shall we say the wisdom of the crowd that brought this forth.
There is no viable indication that the generative AI of today harbors a hidden overpowering preoccupation with the Roman Empire.
In preparation for writing this column, I tried all kinds of prompts and probes to see if there were some secret internal machinations within several of the popular generative AI apps that might showcase some outstretched Roman Empire considerations. Couldn’t find it. Doubtful that it exists. Of course, for those of you who love conspiracy theories, maybe the Roman Empire is cloaked and behind triple doors of security inside generative AI, unable to be seen or detected without a special underground decoder ring.
If you want to prove to your friends that generative AI is preoccupied with the Roman Empire, you can easily pull the wool over their eyes. All you need to do is enter prompts about the Roman Empire and thus get the generative AI into that realm of context. You can be sneaky and put your prompts into a set of custom instructions that are read by the generative AI when it first launches (see my explanation at the link here), ergo hiding them from the eyes of the person you are scamming.
You might not need to have to go that much trouble anyway, assuming that everyone keeps posting volumes of commentary about the Roman Empire onto the Internet or as queries or prompts into generative AI. We might get our just desserts by our own deeds.
Some final thoughts on this meaty topic for now.
Let’s consider two prominent quotes from those who have a lived experience associated with the Roman Empire:
- Julius Caesar said this: “What we wish, we readily believe, and what we ourselves think, we imagine others think also.”
- Claudius noted: “Say not always what you know, but always know what you say.”
Here’s my point.
I am not especially concerned about the Roman Empire being ingrained into generative AI, though I certainly relish the lessons that can be gleaned from the viral trend that can give us a heads-up on something quite important.
We are heading down a path where generative AI will become increasingly essential to our lives. Oddly or ironically, we will be able to slant or shape the generative AI via our own actions such as by widely and immensely posting content on the web and via whatever we enter as prompts into the AI. Few will realize that we are in the process of making our own bed. All that will occur is a sudden shock and discovery that generative AI has seemingly gone stray. It won’t necessarily be because the AI did so per se, but more so because we fed the beast and shot our own feet in the process.
I wonder what the Roman Empire would have done about generative AI. For all of you out there thinking daily about the Roman Empire, please go ahead and put that thought into your internal preoccupation and let me know what pops out.
As they say, Rome was not built in a day, and nor should generative AI.