Claim: AI can have emotions

Question

It has been said that one of the main differences between human intelligence and artificial intelligence is the ability to feel emotion. But if computers/AI can feel emotion, is there any difference?

Answer 1

1. Brief Summary: The claim “AI can have emotions” is misleading. While AI can recognize emotional expressions through patterns, it does not have emotions. AI is not a conscious human being with unique experiences. The emotional responses that AI gives are based on pattern recognition and being programmed, not genuine feelings or emotions.
2. Primary Sources: https://bmcpsychology.biomedcentral.com/articles/10.1186/s40359-024-01581-4 This study reviewed 44 papers on emotion recognition technology. It found that AI can detect emotions using facial expression, voice, and signals, but emphasized that these are simulations not real experiences. 

https://www.mdpi.com/2075-4418/15/4/456 This explored how Ai uses neuroimaging and machine learning to detect emotions. It is a long read, but summarized it said there was progress but there were lots of real world limitations for AI showing emotions.  
3. Secondary Sources: https://www.forbes.com/sites/bernardmarr/2025/04/23/why-ai-will-never-truly-understand-your-feelings-and-why-that-matters/ This article by Forbes says emotional Ai lacks empathy and ethical understanding, which makes it risky to try to make AI have emotions. 
4. Potential Biases: I think that the scientific journals focus more on technological progress for AI and the future which can hide the ethical concerns of this issue. I also think the Forbes article is focused more on following trends by criticizing emotional AI.
5. Evidence in support of the claim: The evidence I gathered is that AI systems can recognize and simulate emotions using facial recognition, voice, and patterns. 
6. What evidence undermines the claim you are fact-checking? AI lacks consciousness and self awareness which are needed to have real emotions. These “emotional” responses are pre programmed and learned from data. 
7. What happened when you tried contacting the person or group who made the original claim? I have not tried contacting the person who made the claim. 

Answer 2

The statement 'AI can have emotions' is ultimately a matter of personal opinion. In this context, the use of the term 'artificial intelligence,' can be assumed to refer to large language models trained by labs such as OpenAI, Anthropic, and Google Deepmind. On a very basic level, these LLMs work by repetitively predicting the next word. However, this picture leads to misunderstandings about the way these models operate. Research at Anthropic has shown that their LLMs share concepts across languages and can plan ahead in their outputs (e.g., planning a rhyme). Training these models involves feeding them massive quantities of data and letting semantic associations between tokens (often words) form. This is why Joshua Batson, a research scientist at Anthropic claims that "They almost grow organically... There are insane things that these models learn to do, but we don’t know how that happened because we didn’t go in there and set the knobs.” 

Recent research has shown LLMs are capable of simulating consistent emotions. However, these 'emotions' are referred to as 'simulated emotions,' aligning with the view that they aren't genuine emotions. 

In the end, questions about LLMs' emotions are answered by personal views on consciousness and emotion. There is no test that can be performed to assess the nebulous concept of 'true emotion' in LLMs. Much of the evidence of emotion in animals comes from observing behavior and reaction. Observing the same behavior in LLMs as we do in animals hasn't lead to widespread belief in machine emotion. This may be due to their temporal difference and lack of clear, unified physical body. 

Sources:
https://www.ibm.com/think/topics/large-language-models
https://www.anthropic.com/research/tracing-thoughts-language-model
https://www.technologyreview.com/2025/03/27/1113916/anthropic-can-now-track-the-bizarre-inner-workings-of-a-large-language-model/
https://arxiv.org/abs/2504.14706
https://online.uwa.edu/news/empathy-in-animals/
https://www.barna.com/research/emotional-ties-ai/