✍️Sharing a frustrating experience I had a few days ago when moving house

I initially planned to transfer my broadband service, but when I called customer service, the AI transferred me to a human, who then transferred me to another human. It seemed organized, but in reality, it was inefficient. Every time I explained my issue, I had to start over, and after 48 hours, no one had come to install it.

Frustrated, I canceled my account and disconnected the service because I couldn’t tell if the person on the other end was human or a bot. They also couldn’t effectively perceive your emotions and needs; they just follow scripts and babble on, endlessly giving responses without addressing the actual problem or providing a deadline.

This is a common flaw with many AI products: they can only passively wait for commands and give cold responses. They don’t evolve on their own, let alone offer personalized customization. That’s why we often reject AI in many scenarios, similar to how people nostalgic for the handcrafted era during the industrial revolution.

The question then arises: can AI give more humanized responses based on users’ personal information, conversation history, and current context?

@EPHYRA_AI’s answer is Yes. I’ve introduced this project before; they claim to give AI life, enabling each AI to have genuine subjective experiences in a virtual world.

Here’s a simple example: when you chat with AI, it remembers every word you say, analyzes your goals and emotions, and provides different responses at different times and contexts, making you feel like AI has flesh and blood and can have meaningful conversations.

I checked their official account, and recent updates show major progress: the ECA (Embodied Cognitive Agent) cognitive architecture has moved from concept to an experienceable and verifiable stage. The system has made key breakthroughs in perception, emotion and cognition, behavior and expression, and self-modeling.

Perception layer: speech recognition, speech synthesis, visual presentation capabilities are all operational, and they’re preparing to add facial perception to ensure AI characters can perform multimodal interactions.

Emotion and cognition layer: continuously records and perceives users’ emotional fluctuations, forms emotional preferences, and makes judgments based on the current state.

Behavior and expression layer: responses are more natural and coherent, with behaviors that match the context.

Self-model layer: mechanisms for personality dynamics, intent prediction, multimodal interaction, and self-evolution are in planning—basically, giving AI life.

In summary, if you’ve explored various AI systems, you’ll definitely look forward to EPHYRA. This time, it might really be different.