Ai Vs Therapist? Which Should I choose?

An AI therapy chatbot can genuinely ease depression symptoms in the short term, but the current evidence shows that benefit fades within months and barely touches anxiety at all, which makes it a supplement to therapy rather than a stand in for it.

Millions of People Are Already Telling a Chatbot What They Would Not Tell a Person

That is not really surprising. An app is free, it is there at two in the morning, and it never looks tired of hearing from you. The pull makes sense, the question is whether it does anything close to what a therapist does, and that turns out to have a fairly specific, evidence backed answer rather than a simple yes or no.

The Short Term Effect on Depression Is Real, and Then It Shrinks

A 2024 systematic review and meta-analysis in the Journal of Affective Disorders pooled 18 randomised controlled trials covering 3,477 people and found meaningful improvements in both depression and anxiety after AI chatbot use, strongest at around eight weeks, but with no significant effect remaining at the three month follow-up (Zhong et al., 2024, DOI: 10.1016/j.jad.2024.04.057). The pattern is consistent: a real lift, on a short clock.

Anxiety Responds Less Than Depression Does

A 2025 meta-analysis in the Journal of Medical Internet Research, focused specifically on generative AI chatbots, pooled 14 randomised trials with just over 6,300 participants and found a statistically significant but fairly small overall effect on negative mental health symptoms, with social style chatbots outperforming task focused ones (Zhang et al., 2025, DOI: 10.2196/78238). Anxiety in particular tends to be the harder symptom for a chatbot to shift, which lines up with what counsellors see clinically, anxiety usually needs a relationship that can push back, not just a responsive script.

The Evidence Base Is Still Genuinely Thin

A 2024 systematic review in JMIR Mental Health looking at self-administered interventions built on natural language processing found meaningful reductions in both depressive and anxiety symptoms across 21 studies, but rated the overall certainty of that evidence as very low, largely due to bias and inconsistency between trials (Villarreal-Zegarra et al., 2024, DOI: 10.2196/59560). None of this means chatbots are useless, it means the research hasn't caught up to the hype yet, and claims of chatbots matching or beating human therapists should be read with real caution.

Where a Chatbot Genuinely Earns Its Place

The honest use case sitting underneath all three studies is support between sessions, not instead of them: mood tracking, journalling prompts, a place to name a feeling at 2am before bringing it to your actual counsellor. Naming a feeling out loud, even to an app, is a real cognitive tool, it turns something vague into something you can work with. What it cannot do is build the kind of relationship where someone who knows your history gently challenges a pattern you can't see in yourself.

AI Chatbot vs In-Person Therapy

FactorAI chatbotIn-person or online therapy with a counsellorAvailability24/7, instantScheduled sessionsDepression symptomsReal short-term improvement, fades by 3 monthsSustained improvement with ongoing workAnxiety symptomsSmall, inconsistent effectEstablished, evidence-backed treatmentBest used forBetween-session support, journallingThe actual clinical work

Related reading

• Anxiety
• Depression
• 'Finding a Therapist Near Me'? The best first step in your mental health journey

Have you ever told an app something you would not have said to a person yet?

If a chatbot has become your main outlet, that is worth bringing into the room with an actual counsellor rather than carrying alone. Our blog covers more on using digital tools alongside therapy, and articles by our counsellors go deeper on building support that actually holds.

Are you considering starting therapy? Find your perfect counsellor today. Find your perfect counsellor today.