Videos or conversation: is there really a difference?

The same screen led to different results depending on whether someone responded right away to what the child did. We gathered studies on video calls, background TV and watching together to set out how watching videos and having a conversation differ for children aged 0–5. This is not an article about getting rid of screens. It is about turning time in front of them into conversation.

The difference researchers found was not the screen but the response

Two-year-olds learned new words exchanged live over a video call, but not when they watched the same scene as a recording.¹ What mattered was not whether there was a screen but whether the other person responded right away to the child. Babies around their first birthday also remembered a person and learned new things over live video calls.²

In one experiment, 36 children aged 24–30 months were split across three conditions: meeting in person, a video call, and a recording of the same scene. They learned new verbs only in the two conditions where the partner reacted in real time.¹ In another, babies aged 12–25 months saw either live video calls or pre-recorded videos over a week. Only the babies in the live condition recognised and preferred the person a week later, and the oldest children also learned more new words.²

Older research points the same way. When nine-month-old American babies met a native Chinese speaker twelve times in person, they kept the ability to tell Chinese sounds apart, but the same person saying the same things on video or on an audio recording had no effect.³ Babies learned the same words better from a person sitting across from them. Keep in mind that in all three studies, the one responding was a person.

A video left on means less conversation at home

A study recording 220 Australian families from 12 to 36 months found that more screen time went with fewer adult words, fewer child vocalisations and fewer back-and-forth exchanges.⁴ A TV left on with no one watching did the same. In a lab living room, adult TV playing in the background reduced both the amount and quality of parent–child interaction.⁵

The Australian study recorded the sounds of each home for 16 hours a day, every six months. At 36 months, each minute of screen time went with 6.6 fewer adult words, 4.9 fewer child vocalisations and 1.1 fewer conversational turns.⁴ The background TV experiment followed 51 children aged 12, 24 and 36 months playing with a parent for an hour, with adult TV on for 30 minutes of it.⁵

None of this means screens ruin children. These are observational findings, so they cannot show cause, and the size of the effect differs from home to home. What they do suggest is one thing: for as long as a screen is on, there may be less room for back-and-forth talk.

Should screens go completely? Watching together was different

Not necessarily. A meta-analysis of 42 studies with 18,905 children linked more screen time with lower language scores, but watching with an adult and watching educational programmes pointed the other way.⁶ The American Academy of Pediatrics (AAP) also favours family rules that focus on what children watch and with whom, over rules that only count minutes.⁷

The correlations in the meta-analysis were small. Screen time and language had an r of −0.14 and background TV −0.19, while co-viewing (r 0.16) and educational programmes (r 0.13) went with better language.⁶ The point is that the direction flips. The same screen becomes a different experience when an adult adds words alongside it.

In its 2016 recommendations the AAP stressed parents and children using media together, and its more recent guidance says it does not set one time limit for every child. It explains that family rules built around balance, content, watching together and conversation are linked to better outcomes than rules that only track time.⁷

Recommendations by age at a glance

The World Health Organization advises no screen time under one, and no more than an hour a day at two and at three to four, with less being better.⁸ The Korean Pediatric Society advises avoiding digital media until two, then under an hour a day, watched together. The AAP advises avoiding screens under 18 months, except video calls with family.⁷⁹

• Around the first birthday: the World Health Organization advises no sedentary screen time for babies under one and for one-year-olds.⁸ The AAP makes video calls with family the only exception under 18 months.⁷
• Up to two: the Korean Pediatric Society advises avoiding digital media until age two.⁹ The AAP suggests that at 18–24 months, children watch good-quality programmes together with an adult.⁷
• Age two: the World Health Organization says no more than an hour a day, and less is better.⁸
• Ages three to four: the World Health Organization says no more than an hour a day.⁸ The AAP and the Korean Pediatric Society say under an hour a day, watched together, for ages 2–5.⁷⁹
• All ages: the World Health Organization recommends that when children are sitting still, a caregiver reads and talks with them.⁸

The video-call exception is about calls with a person who responds in real time. Talking with an AI is not an exception in any of these recommendations.

Five ways to turn video time into conversation

Rather than removing screens, it is more realistic to fit conversation in around them: pause and ask while you watch together, talk about it later, turn off background TV, let your child choose, and agree on the end in advance. None of these needs anything, and all five bring the responding and back-and-forth that research highlights into screen time.

1. Pause and ask while you watch together. Stop and ask, "Where do you think that puppy is going?" Then wait for your child's answer. A sound or a gesture counts as an answer too.
2. Talk about it later. At mealtime: "Remember that rabbit? What was it looking for?" It is practice in talking about things that are not in front of you.
3. Turn off background TV. When nobody is watching, switch it off. Just having it on reduced back-and-forth between parents and children.⁵
4. Let your child choose. Offer two options, and ask why they picked the one they did.
5. Agree on the end in advance. Say "When this finishes, we turn it off" together before you start. The last of the serve-and-return steps described by Harvard's Center on the Developing Child is also about noticing endings and beginnings.¹⁰

What did research on talking characters show?

When the characters in a cartoon asked questions and reacted to children's answers, 4- to 6-year-olds talked more, understood the content better, and actually spent less time just staring at the screen.¹¹ But this was a short experiment with a single video, and it does not mean a responsive character can stand in for conversation with a person.

In this experiment, researchers added conversation to a public-television science cartoon using natural language processing, and randomly assigned 77 children aged 4–6. Children who watched the version with conversation understood the science concepts better, talked more and showed more positive emotion.¹¹ In another experiment by the same team, 117 children aged 3–6 were randomly assigned, and those who read with an AI partner that asked questions understood the story as well as those who read while talking with a person.¹² That agent, however, followed a pre-written script, and the results were measured right after a single reading.

A 2025 experiment paints a slightly different picture. Sixty-seven children aged 5–8 learning English as a foreign language were randomly assigned. Those who read with an agent built on a recent language model scored higher on comprehension than those who read with a parent, but emotional engagement and talk about the story were higher with a parent.¹³ No study has yet looked at the effect of AI conversation on babies under two. Several studies show the value of a response, but research on AI conversation is still at an early stage.

How we built Alphabebe

Instead of playing videos, Bebe is designed to ask your child questions, wait, and pick up on what they say. There is no endless video or feed, there is a daily limit, and Bebe goes to sleep at night too. Most of the studies above looked at conversation with people, so Bebe does not replace conversation with parents.

It adds one more thing to talk about with your child, and keeps today's conversations in the parent space for you. For babies under two, we only suggest using it with a caregiver holding the screen, reading and playing together. About Alphabebe

Sources

1. Roseberry S, Hirsh-Pasek K, Golinkoff RM. 2014, Child Development 85(3):956–970. DOI 10.1111/cdev.12166
2. Myers LJ, LeWitt RB, Gallo RE, Maselli NM. 2017, Developmental Science 20(4):e12430. DOI 10.1111/desc.12430
3. Kuhl PK, Tsao F-M, Liu H-M. 2003, PNAS 100(15):9096–9101. DOI 10.1073/pnas.1532872100
4. Brushe ME, Haag DG, Melhuish EC, Reilly S, Gregory T. 2024, JAMA Pediatrics 178(4):369–375. DOI 10.1001/jamapediatrics.2023.6790
5. Kirkorian HL, Pempek TA, Murphy LA, Schmidt ME, Anderson DR. 2009, Child Development 80(5):1350–1359. DOI 10.1111/j.1467-8624.2009.01337.x
6. Madigan S, McArthur BA, Anhorn C, Eirich R, Christakis DA. 2020, JAMA Pediatrics 174(7):665–675. DOI 10.1001/jamapediatrics.2020.0327
7. AAP Council on Communications and Media. “Media and Young Minds.” 2016, Pediatrics 138(5):e20162591. DOI 10.1542/peds.2016-2591 · American Academy of Pediatrics screen time guidance (aap.org)
8. World Health Organization. “To grow up healthy, children need to sit less and play more.” 2019-04-24, who.int
9. Korean Pediatric Society, recommendations on digital media use for infants and young children (in Korean)
10. Center on the Developing Child at Harvard University. “Serve and Return” · “5 Steps for Brain-Building Serve and Return.” developingchild.harvard.edu
11. Xu Y, Vigil V, Bustamante AS, Warschauer M. 2022, Journal of Applied Developmental Psychology 81:101439. DOI 10.1016/j.appdev.2022.101439
12. Xu Y, Aubele J, Vigil V, Bustamante AS, Kim YS, Warschauer M. 2022, Child Development 93(2):e149–e167. DOI 10.1111/cdev.13708
13. Xiao F, Zou EW, Lin J, Li Z, Yang D. 2025, British Journal of Educational Technology. DOI 10.1111/bjet.13615

This article is not a diagnosis or a treatment. If you are worried, talk to your paediatrician.