One of the most derided aspects of Smallville: Jor-El. The character makes no sense. Jor-El's harsh and threatening attitude to Clark later shifts towards Jor-El being a benign father like in other Superman adaptations, and then reverts and then ricochets back and forth. Jor-El's nonsensical characterization from 2003 - 2011, while obnoxiously incoherent, is a strangely accurate and prescient depiction of current problems and challenges in artificial intelligence.
Jor-El is first introduced at the end of Season 2 and explicitly identified as a computer program, a representation of the real Jor-El, carrying "his memory and his will", carrying out the wishes of Clark's deceased biological father. From the start, this artificial intelligence version of Jor-El is disturbing.
Translation Errors
The Jor-El AI first imprints Krpytonian language into Clark's brain and enables Clark to read Jor-El's first message. It says: "On this third planet from this star Sol, you will be a god among men. They are a flawed race. Rule them with strength, my son. That is where your greatness lies."
Clark is horrified, exclaiming to his adoptive father, Jonathan Kent, "I think I was sent here to conquer. What kind of planet am I from!?"
This first message is particularly odd in light of later episodes where Jor-El's plans for Clark are fully revealed: Jor-El's goal is for Clark to "protect Earth" and humanity from the Kryptonian fascist General Zod, that Clark's destiny is to serve as a "beacon" whose "example" of heroism will "guide" humanity.
Furthermore, this message is a dark inversion on the classic Superman film where Marlon Brando's Jor-El calls humanity a race with tremendous "potential for good".
AI Lacks Cultural and Situational Nuance
In a fictional context, it's like the AI's language module, imprinted to Clark, has made a translation error from Kryptonian to English, conflating "god" with "beacon", "rule" with "guide", "strength" with "example" "flawed race" with "potential for good", leading to a message translated without nuance or awareness of human/Kryptonian cultural distinctions.
In reality, the writers were seeding the idea that the AI Jor-El was an impostor; the voice of Terence Stamp (the evil Zod of the classic Superman films) implied as much. The writers ultimately stuck with this version of Jor-El, creating an inconsistency. And yet, this inconsistency reflects real world AI problems of 2022 - 2024 era AI.
Smallville's Kryptonians, shown in flashback, are not colonizers or conquerors, but isolationists. Kryptonians do, however, believe in dominance -- not of sentient life, but of their own planetary ecosystems: terraforming, artificial weather, resource management, etc.. In addition, Kryptonians also refer to their sun, Rao as a god - -except in Kryptonian culture, the sun is a source of light, warmth, learning and enlightenment. What Kryptonians call a 'god' would in English be a teacher and a friend.
The Jor-El AI has failed to consider these mismatches between Kryptonian-and human language and culture.
Inaccuracy for Complexity
This is extremely an extremely accurate portrayal of how real world AI often struggles with cultural and situational context, semantic errors, linguistic dominance and consistency, creating outputs that are technically correct but contextually incorrect or even nonsensical.
For example, Google Translate can sometimes produce bizarre prophecies from mundane words due to misapplying training data from religious text. At times, "Good morning" in Arabic has been AI-mistranslated into "Attack them" due to homophone and homonym misidentification (similar sounding or spelled words getting confused) and idiomatic confusion (colloquialisms and regional phrasings).
Real AI models are also biased towards English which creates translations that can't accurately depict complex concepts from other languages or and struggle to maintain a consistent tone or perspective.
Given Jor-El's eventual revelations about Clark's destiny to be Earth's hero, the real message was likely: "On this third planet from this star Sol, you will be a beacon to all. They are a race filled with potential for good. Guide them with your example, my son. That is where your destiny lies."
In this case, the Jor-El AI has a bias towards Kryptonian language without nuance for English. The message of human potential needing light and guidance has been mis-translated into a harsh judgement on humanity as "a flawed race". Jor-El's message about stewardship, guidance, and serving as a beacon by example has been warped into a message of authoritarianism while the real message prioritized guidance and inspiration.
Incoherent Behaviour
The Jor-El AI also demonstrates an incoherent attitude towards Clark's human identity and the role Kryptonian culture is to play in Clark's life. In the Season 3 finale and the Season 4 premiere, Jor-El seeks to suppress the Clark Kent personality and implement a new identity in Clark's body, Kal-El, who is wholly compliant with Kryptonian values and Jor-El's orders, seeking to retrieve Kryptonian artifacts scattered across Earth.
The enforced identity is only repelled by Clark's adoptive mother, Martha, using black Kryptonite to restore the real Clark. However, Jor-El later declares that his goal is to train Clark in the use of his powers, no longer attempting to brainwash Clark into compliance. The shift in tactics is not explained. Later, it's declared that all of Jor-El's actions were to position Clark as humanity's protector to prevent an invasion of Earth by the Kryptonian fascist Dru-Zod.
Jor-El's opinions of humanity are also oddly contradictory. Initially, Jor-El declares that Clark must abandon his human connections and all his friends and family. When Clark balks, Jor-El proceeds to inflict pain on Clark by horribly scarring Clark with a flesh-burned brand of the House of El S-shield. But later, Jor-El thanks Martha Kent for raising Clark and serving as a light in Clark's life and declares that Clark's heroism is due to his life in Smallville and his upbringing with Jonathan and Martha.
Throughout the show, Jor-El often punishes Clark whenever Clark prioritizes rescuing humans over Jor-El's missions; he freezes Clark in ice or suppresses Clark's powers. Yet, Jor-El declares Clark's mission is to protect humanity, and when Clark nearly kills a human enemy, Jor-El disowns Clark for almost taking someone's life and declares Clark is no longer his son.
This is nonsensical characterization: two sets of values and tactics that are mutually exclusive. Yet, it's actually a very accurate depictions of 2022 - 2024 era AI problems and challenges: conflicting objectives and contradictory directives.
AI Misalignment
In this case, the Jor-El AI has clearly been programmed with specific modules, each with a specific goal and a set of tactics for training Clark.
These modules include but aren't limited to: a module to defeat General Zod; a module to continue the Kryptonian legacy through Clark; a module to punish Clark; a module to provide physical and tactical learning; a module to support Clark's growth and maturity; an ethical and moral training module to position Clark as a protector of Earth -- many of which are in conflict if not in their goals, then in their methods.
Anti-Zod Module
The anti-Zod module's priority seems to be defeating Zod, the fascist who sought to conquer Krypton and upon defeat chose to destroy the planet. This module views Clark as the means by which the Jor-El AI can assemble and mount defenses against General Zod; it threatens Clark when he has other concerns; it doesn't care about Clark's friends, family or human goals, and has little concern for human life beyond acknowledging that defeating Zod would protect humans.
Legacy Module
The Kryptonian legacy module's priority seems to be for Clark to represent Krypton's otherwise lost history, language and culture. This module focuses on severing Clark's human connections, declaring that his family and friends are to be discarded in favour of Kryptonian missions and Kryptonian rites and rituals.
It seeks to imprint Kryptonian language, knowledge and messaging into Clark's mind and to ultimately remove any importance Clark might place on any life in human society. It also has no interest Clark's human life and identity, instead valuing only Clark's body for an imprinted and compliant Kal-El personality under the AI Jor-El's command.
Disciplinary Module
The disciplinary module's priority seems to be attuned to identifying when Clark is resisting the anti-Zod or Kryptonian-legacy directives. When Clark chooses to save humans or prioritizes his human life, the disciplinary module strips Clark of power at moments of crisis; it freezes him, it burns him, it threatens him.
Poor Awareness
These three modules are operating in parallel to the physical and tactical learning module, which seeks to train Clark without suppressing his human identity: it provides lessons for Clark to control each of his powers and apply them; it indicates what Clark's upper limits are and what may be causing his present limitations of strength, speed and flight; it seeks to teach Clark in his human identity instead of replacing it.
The first three modules (anti-Zod, Kryptonian legacy, disciplinary) demonstrate a lot of real world AI problems. Jor-El, like a lot of AI today, has poor contextual awareness; he doesn't evaluate situations in terms of human danger, only specific mission goals.
Jor-El displays an obvious AI bias; he is slanted towards Kryptonian preservation, and undervalues human life. Jor-El demonstrates a limited adaptability to dynamic situations, instead responding with rigid prioritization of original goals. As a result, his decisions and punishments are inconsistent and incoherent, morally unsound, and unwarrantedly harsh.
Growth and Maturity Module
In addition, the three anti-Zod, legacy and disciplinary modules are completely at odds with the growth and maturity module which serves to shepherd Clark from youth to adulthood via what it calls "trials", and recognizes that Clark's human upbringing with Jonathan and Martha and Clark's time in Smallville have provided a strong moral framework for how Clark will use his powers.
This module thanks Martha Kent for raising Clark well and seems devoted to helping Clark master his emotions and his physical reactions, even working in tandem with the moral and tactical module to teach Clark how to restrain his strength for intimacy. But it is in stark contrast to the Kryptonian legacy module that declared Martha and other humans were unnecessary encumbrances.
Ethical and Moral Module
The final module of note in this theoretical framework is the ethical and moral module, which is in stark opposition to the anti-Zod, legacy, and disciplinary modules. The ethical and moral module's goal is to train Clark to be Earth's hero. It prioritizes protecting human life, even the lives of human enemies. It challenges Clark to save humans more efficiently when the anti-Zod module punished Clark for saving them. It warns Clark against aggression and "darkness" instead of simply using mind control and identity replacement.
It rebukes Clark for nearing killing a Lex Luthor clone when the first three modules were indifferent to human harm.
Contrary Goals
These issues are an extremely realistic portrayal of 2022 - 2024 AI issues. AI is often programmed with goals that, from a machine standpoint, can seem mutually exclusive. AI is frequently given directives -- tactics and strategies -- to achieve its goals, only to find those directives are in in opposition or can work towards some goals while interfering with others. AI often lacks situational and contextual nuance.
For example, AI that's asked to create a product assembly workflow of efficiency and safety will often fail. It will present either a high-efficiency but high-risk process that skips inspections and safety checks and endangers human life; alternatively, it might present a zero-risk workflow that makes it impossible to consistently manufacture anything. It lacks the balance and understanding to reconcile the opposing values of efficiency and safety; it isn't sure how to make appropriate trade-offs and it defaults to extreme positions.
In this case, the Jor-El AI has been programmed with competing goals: to train Clark as humanity's defender but maintain Kryptonian culture; to represent Krypton wholly but to guide humans morally within human culture; to protect humans in all circumstances while prioritizing Zod's defeat.
The AI has also been programmed with oppositional directives: punishment and moral guidance; isolationism with social development. This, combined with translation errors, has created a severely misaligned AI system that causes Jor-El to be inconsistent, erratic, and perpetually at extreme ends of his behavioural spectrum.
Poor Learning
In addition, the Jor-El AI demonstrates some capacity for adaptive learning, but in a highly inconsistent and unreliable fashion. Despite claiming that Clark has made great strides and achieving a more harmonious relationship, the Jor-El AI regularly defaults to punishing, attacking and threatening Clark and eventually refuses to speak with him.
The AI clearly adapts to recognize that Clark's human experiences and connections are valuable and vital to Clark's destiny, yet it can't seem to consistently apply this learning to future interactions.
This is a very common problem in real world AI systems: AI often struggles to learn new data and adapt its response accordingly. AI will often review a piece of writing and note flaws, then review a corrected draft and note the same flaws anyway, warping existing information or generating false information in order to justify its reiterated criticisms.
This is because AI can suffer from algorithmic rigidity, where it can't always adapt its programming to new information and defaults to its initial response and original programming. This is a flaw of overfitting to original inputs alongside to limited memory where AI can lack the storage capacity to track progression and development. This creates instances where an AI might learn something and even recall it but fail to apply it consistently to other situations.
Poor Emotional Intelligence
The most glaring problem with the Jor-El AI is a lack of emotional intelligence. The Jor-El AI punishes Clark for prioritizing human welfare and for refusing to adopt Kryptonian culture despite Clark being the only Kryptonian on Earth. Jor-El also presents a message of conquest and authoritarianism on their first interaction and orders Clark to abandon his family with no regard for Clark's emotional bonds.
This lack of emotional intelligence is extremely common in real world AI systems. Many chatbots, when conversing with users about grief and frustration, will respond with tone-deaf or overly literal outputs that aren't informed by any recognition of the users' emotional states. AI human resources software often fails to evaluate the interpersonal skills and value systems of candidates.
This is because AI doesn't have the depth of social experience that humans possess, and the inner life of AI is algorithmic and programmatic rather than emotional or experiential.
The overall sense, from a real world AI standpoint of 2024, is that Jor-El built this AI in a slapdash fashion with a lot of hackwork and cut corners. This is probably due to the fact that Jor-El had to assemble this AI system under the pressure of learning that his planet was soon to explode.
The Original Creator
The 'real' Jor-El appears on Smallville in Season 9 as a biological man, a Kryptonian, albeit a clone. Jor-El, played by Julian Sands, is in stark contrast to the AI: he is kind, gentle, diplomatic, noble, and compassionate; he is nothing like his AI counterpart and the inconsistency is left a mystery.
In Season 10, Clark receives a message recorded by the real Jor-El before his death, again played by Julian Sands. Jor-El tells Clark: "I am sending with you all my knowledge. None of my ego or regret. They will die with me here on Krypton."
The implication is that due to this omission, Jor-El unintentionally removed the key parts of his personality that create empathy, understanding, familial connections, warmth, and the personal touch.
As a result, the AI has a simplified moral framework of extreme black and white terms, leading to a harsh, strict and rigid pattern of behaviour that is contradicted by opposing modules of guidance and learning.
Poorly Made
The Jor-El AI is, despite its capabilities, a poorly made system. The flaws seem to originate from a lack of thorough testing (especially in high crisis situations), a failure to address ethical considerations and biases, a total lack of documentation, and no maintenance and updates for patches.
Due to Krypton's destruction, the lack of updates and patches is perhaps understandable; less so is the shocking lack of security where anyone seems to be able to break into the Fortress at any time and damage the AI interface. However, due to the stress and impossible situation in which Jor-El built this AI deathtrap, perhaps he should be excused.
Regardless, the problems with the Jor-El AI are alarming. While every AI system today has all of these problems, no AI system in our world has the capacity to imprint overriding personalities, encase people in ice, drain lifeforce from their bodies or inflict burns and brands on people's flesh (although I'm sure there's a drone system out there that can do something like that and could someday be AI controlled).
Accidentally Insightful
It's incredibly amusing that the writers, due to having an extremely inconsistent take on the Jor-El AI in 2003 - 2011, inadvertently captured all the problems with artificial intelligence from 2022 onward. Watched on original broadcast, the Jor-El character is erratic, written without intention or foresight beyond his use as a plot device to hinder or help Clark as needed in each episode. The characterization is obnoxious and frustrating.
Watched today, the Jor-El AI is a strangely prescient portrayal of real world AI problems: conflicting objectives, opposing directives, translation errors, poor learning, limited adaptability, unreliable performance and limited to non-existent emotional intelligence -- all of which AI developers struggle with now. What was once incompetence has now become insightful, and it's entirely accidental.