Can AI model look and describe like a 4 year old?

Can AI models now look at an image and output speech like a 4-year-old finally?

References:
https://arxiv.org/pdf/2503.15633