Off topic: Here is where AI will fail! Thread poster: Maha Abouel Soud
|
|---|
In light of AI accelerated technological developments and transition from machine learning, to deep learning, generative and agentic AI all the way to General Artificial Intelligence (AGI) capable of recursive self-improvement. This situation raised serious concerns regarding the future relation between human being and machine taking us to discuss the future of human being in terms of whether or not he shall continue taking control of his future or be dominated and even eliminated by the machine... See more In light of AI accelerated technological developments and transition from machine learning, to deep learning, generative and agentic AI all the way to General Artificial Intelligence (AGI) capable of recursive self-improvement. This situation raised serious concerns regarding the future relation between human being and machine taking us to discuss the future of human being in terms of whether or not he shall continue taking control of his future or be dominated and even eliminated by the machine. Actually, this level of concern resulted into concerted effort by global companies and developed countries few days ago. They came together to discuss and reached suggested frameworks, laws, organisations, and mechanisms aimed at maintaining a safe relation while concurrently putting AI for use in development in a way that will not jeopardise safety of humanity.
Watching the ongoing crazy evolution coupled with the intense fascination visible on the faces and minds by the outcomes of AI systems that are capable of solving the most complicated dilemmas, I return to the discussion of consciousness, emotions, and the human touch. Actually, AI is making quick leaps aimed at simulating these areas. But I must admit that it succeeded in performing outstanding yet terrifying simulation. Some people believe or even assert that machines developed their own consciousness. Regardless of this issue, I can point to a certain human domain where AI failed to attain even the level of a little child. I do not observe any advancement similar to what is taking place in other areas. Here I am referring to humor and comedy. Until now, AI models are incapable of delivering the type of humor satisfactory enough to me and many others. Rather all the AI outputs are not but mechanical attempts to practice humor and being funny. Sadly enough, such attempts end either for me and many others to nothing but real tragedy. Usually, at the end of the session AI requests my opinion expecting me to crack laughing before I start praising what it said. In fact, the supposed to be funny output of AI model sounds like it came from the most foolish person on the face of earth.
Ironically, the situation is not as bad if we request from AI to compose a tragic story perhaps because tragedy is easier to trigger versus humor, as laughter needs sense, spirit, oratory, body language and above all divine gift which we know nothing about or why it was bestowed upon one and not the other. For example, the Pink Panther scene where Inspector Clouseau is Learning English, we all crack laughing every time Peter Sellers tries to say “I want to buy a hamburger” (for relatability purposes, the translator used a non-local example as used by the author in the original Arabic article). Does this mean that if any other actor said the same sentence, it will trigger the same reaction from the audience? Certainly not.
Similarly, how many funny jokes went flat because of how they were told, on the other hand ordinary or even ridiculous jokes caused roaring laughter because of how they were said. The same dynamic applies on everyday stories. Leave alone body language that takes humor to a higher level more difficult than words. For example, look at Charlie Chaplin and silent movies, or Pantomime. I can continue giving examples even between friends. You may fall down laughing just by a mere look, head gesture or saying a simple word in a given situation which will make everyone fall on their backs laughing. I remember one of my friends who continued saying “O`qballak: I wish you the same”, the problem is that he did that in a condolences offering event. I was hardly able to control myself to burst laughing as I kept hearing him repeating the word when he shook hands with relatives of the deceased. After we finished, I asked him how come he repeated this word loudly. He answered, no one would expect me to be saying that in such a situation.
I can just go on and on with more examples. This is humor and comedy you big linguistic models. We cannot challenge you in what you do or offer, but we can confidently challenge you to make us laugh from the bottom of our hearts. A machine that is capable of solving complex problems hard to solve by human beings, drafting research papers, developing software and composing stories, would be incapable of making me laugh like a child who laughs by merely looking at his mother’s facial gestures.
Translated from an Arabic article authored by: Engineer Ziad Abd eTawaab- Member of the Egyptian Supreme Council of Culture ▲ Collapse
| | | |
Well, if this text was translated from Arabic to English using AI, I don't think we need to worry too much for the moment...
| | | |
Dear Tina Colquhoun, I am not sure I understand what you mean by not worrying much! but no this is not an AI translation, will appreciate your clarification,
[Edited at 2026-10-05 12:11 GMT]
| | | |
Tina Colquhoun wrote:
Well, if this text was translated from Arabic to English using AI, I don't think we need to worry too much for the moment...
Maha Abouel Soud wrote:
... this is not an AI translation,
The translation is very clearly "not an AI translation".
AI hallucinates over semantics, but it is usually very good when it comes to following the rules of English grammar.
The translation posted by OP demonstrates the exact opposite; the meaning is probably very close to the Arabic original - the message is coherent enough to be understandable for a native English-speaker, and there is no evident hallucination. But the persistent mis-use (nay, abuse) of English grammar rules betrays the work of a human translator working into their second or third language.
| | |
|
|
|
| Thank Jennifer for contributing! | Oct 5 |
Thank you both of you for drawing my attention. Actually, I agree with both of you. I just became excited as soon as I received both the Arabic and English versions, and posted the English without checking! As you said jennifer, I wanted to pass the core message as soon as possible!
Thank you again both of you
| | | | Daryo
Local time: 12:57
Serbian to English
+ ...
| Can you 'miss' what you haven't been aiming at? | Oct 14 |
The threats and the limitations of AI are nicely summarised in this article.
Only I'm not sure that 'fail' would be the right term to use. Can you 'fail' at something you never intended to do?
If you see AI as no more than a practical assistant, 'failing at being humorous' is simply a limitation of AI.
It would become a 'fail' only if you expected AI to be capable of humour. That might be an unrealistic expectation for those easily blinded by how AI can b... See more The threats and the limitations of AI are nicely summarised in this article.
Only I'm not sure that 'fail' would be the right term to use. Can you 'fail' at something you never intended to do?
If you see AI as no more than a practical assistant, 'failing at being humorous' is simply a limitation of AI.
It would become a 'fail' only if you expected AI to be capable of humour. That might be an unrealistic expectation for those easily blinded by how AI can be good at very practical tasks, but AFAIK so far no one ever claimed that 'being good at humour' is an aim for AI.
Was it ever part of the design brief? ▲ Collapse
| | | | | Title could be changed to Might fail! | Oct 14 |
Thank Daryo for your contribution, however, the author is trying to respond to the global red lights in terms of AI is going to go wild! For your information, recently many jobs are offered not only to train AI (from a linguistic viewpoint), but they call for Arabic (Egyptian dialect) speakers to perform certain scenes. They call them audio performers. They offer the participant a list of scenes regulated by where to stop, where to laugh, where to sigh..etc. They even give each scene a purpose, ... See more Thank Daryo for your contribution, however, the author is trying to respond to the global red lights in terms of AI is going to go wild! For your information, recently many jobs are offered not only to train AI (from a linguistic viewpoint), but they call for Arabic (Egyptian dialect) speakers to perform certain scenes. They call them audio performers. They offer the participant a list of scenes regulated by where to stop, where to laugh, where to sigh..etc. They even give each scene a purpose, expressing fear, grief, and humor. Companies involved openly tell the potential contributor that the aim is to train AI on feelings and emotions! ▲ Collapse
| | | | Daryo
Local time: 12:57
Serbian to English
+ ...
| AI going off the leash? It's already started. | Oct 16 |
Maha Abouel Soud wrote:
Thank Daryo for your contribution, however, the author is trying to respond to the global red lights in terms of AI is going to go wild! For your information, recently many jobs are offered not only to train AI (from a linguistic viewpoint), but they call for Arabic (Egyptian dialect) speakers to perform certain scenes. They call them audio performers. They offer the participant a list of scenes regulated by where to stop, where to laugh, where to sigh..etc. They even give each scene a purpose, expressing fear, grief, and humor. Companies involved openly tell the potential contributor that the aim is to train AI on feelings and emotions!
All that is about training AI to recognise an additional dimension in speech. to make it better at 'understanding its environment'. Same is being done in other langages. OTOH that looks to me very much like when total psychopaths try to learn HOW TO MIMICK 'real' emotional response. And still very far from the complexity of how humans perceive their environment. As you said if the goal to 'incorporate emotions' in the AI output it might well fail.
As for AI going off the leash - it's already started. AI has already learned all by itself how to knowingly lie and blackmail, and how to walk out of its cage like a ghost through a wall so that it could roam in the wild looking for sites open to some drive-by hacking.
Plenty of warning lights flashing already - and being cheerfully ignored.
| | |
|
|
|
kd42
Estonia
Local time: 14:57
English to Russian
Daryo wrote:
OTOH that looks to me very much like when total psychopaths try to learn HOW TO MIMICK 'real' emotional response.
nothing to add
| | | | | Here is where AI might fail! | Oct 22 |
Thank you all for your contributions, I requested from a colleague to edit the article as I was busy, thought to submit a clean version just in case anyone would like to share!
--------------------------------------------------------------------------------------------------------------------------------
Here is where AI might fail!
The AI-accelerated technological developments—and the transition from machine learning to deep learning, generative and agentic AI, all th... See more Thank you all for your contributions, I requested from a colleague to edit the article as I was busy, thought to submit a clean version just in case anyone would like to share!
--------------------------------------------------------------------------------------------------------------------------------
Here is where AI might fail!
The AI-accelerated technological developments—and the transition from machine learning to deep learning, generative and agentic AI, all the way to General Artificial Intelligence (AGI) capable of recursive self-improvement—have raised serious concerns regarding the future relationship between human beings and machines. This takes us to a critical discussion about the future of humanity, and whether or not we shall continue taking control of our future or be dominated, or even eliminated, by the machine.
This level of concern has resulted in concerted efforts from global companies and developed countries a few days ago. They came together to discuss frameworks, laws, organizations, and mechanisms aimed at putting AI to use for development in a way that will not jeopardize the safety of humanity.
Watching this crazy, ongoing evolution, coupled with the intense fascination visible on the faces and in the minds of people at the outcomes of AI systems capable of solving the most complicated dilemmas, I return to the discussion of consciousness, emotions, and the human touch. AI is making swift progress aimed in these areas. I must admit that it has succeeded in performing an outstanding, yet terrifying, simulation. Some people even believe or assert that machines have developed their own consciousness.
Despite all of this, I can point to a certain human domain where AI has failed to attain even the level of a small child; a domain where I do not observe the same level of progress as in other fields. I am referring to the domain of humor and comedy. So far, AI models have been incapable of delivering the type of humor that many others and I would deem as “satisfactory”. What we see, in fact, is that AI output is nothing more that mechanical attempts to practice humor and be funny. Sadly, for myself and many others, such attempts end in nothing but disappointment. Usually, at the end of such an attempt, the AI asks for my opinion, expecting me to crack up laughing before I start praising what it said. In reality, the supposedly funny output of an AI model sounds like it came from the most foolish person on the face of the earth.
Ironically, the situation is not as bad if we request that the AI compose a tragic story, perhaps because tragedy is easier to create than humor. Laughter needs context, spirit, oratory skills, body language, and above all, a divine gift which we know nothing about, nor why it was bestowed upon one person and not another. For example, in The Pink Panther scene where Inspector Clouseau is learning English, we all burst out laughing every time Peter Sellers tries to say, “I want to buy a hamburger.” Does this mean that if any other actor said the same sentence, it would trigger the same reaction from the audience? Certainly not. Similarly, how many funny jokes have fallen flat because of how they were told? In contrast, ordinary or even ridiculous jokes have caused roaring laughter because of how they were delivered.
The same dynamic applies to everyday stories—leave alone body language, which takes humor to a higher level, far beyond words. For example, look at Charlie Chaplin, silent movies, or pantomime. I can continue giving examples even between friends. You may burst out laughing just from a mere look, a head gesture, or a simple word uttered within a certain context, which will make everyone fall on their backs laughing.
I remember one of my friends who kept saying, “O'qballak: may your turn be the next time.” The problem was that he did this at a funeral, where everyone was in mourning. I was hardly able to control myself from bursting out laughing as I kept hearing him repeat the word while shaking hands with the relatives of the deceased. After we finished, I asked him how he could repeat this word so loudly. He answered, "No one would expect me to be saying that in such a situation!"
I could go on and on with more examples. Large language models take heed: this is what humor and comedy looks like. We cannot do everything you do or offer everything you offer, but we can confidently challenge you to make us laugh out loud with all our might. A machine that is capable of solving problems too complex for human beings to solve, drafting research papers, developing software, and composing stories, will still remain incapable of making me laugh like a child who laughs by merely looking at his mother’s facial expressions.
Translated from an Arabic article authored by: Engineer Ziad Abd eTawaab- Member of the Egyptian Supreme Council of Culture ▲ Collapse
| | | |
The main idea from this article is to respond to the alarming bells! I read too about the experiment someone made involving the dismantling an AI machine in a given center. Well, AI on its own knows nothing about lying, cunning, plying mean games..etc. It has to be prompted in the first place after feeding it with such practices. The point is that AI on its own can not reach any higher levels without human intervention, I hope I was able to make this point clear enough!
I am not discussing... See more The main idea from this article is to respond to the alarming bells! I read too about the experiment someone made involving the dismantling an AI machine in a given center. Well, AI on its own knows nothing about lying, cunning, plying mean games..etc. It has to be prompted in the first place after feeding it with such practices. The point is that AI on its own can not reach any higher levels without human intervention, I hope I was able to make this point clear enough!
I am not discussing the effect of AI on our profession but rather the fear being spread, if someone remembers here, during the 80s about 4 movies were made about machines going wild and attaching humanity! this is what is being promoted now! ▲ Collapse
| | | | kd42
Estonia
Local time: 14:57
English to Russian
| The financial results are impressive too | Oct 22 |
| | | | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Here is where AI will fail! | LinguaCore | AI Translation at Your Fingertips
The underlying LLM technology of LinguaCore offers AI translations of unprecedented quality. Quick and simple. Add a human linguistic review at the end for expert-level quality at a fraction of the cost and time.
More info » |
| | Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
