TL;DR: An AI interview is a structured job interview — usually by phone or video — conducted by a voice AI rather than a human. The AI asks the same questions every time, records and transcribes the answers, and scores them against a rubric. Most AI interviews in the UK today are first-round screens — not final decisions. They're faster, more consistent, and when built well, get higher candidate satisfaction than 9-day waits for a human call.

The honest definition

An AI interview is a voice-based conversation with an AI agent. It usually lasts 5–15 minutes. The AI introduces itself, explains it's an AI, asks a set of structured questions tied to the role, listens, and records the answers.

At Lily, every AI interview:

Runs over the phone (not video — care workers apply on mobile, on the go)
Lasts around 5–10 minutes
Asks the same questions in the same order
Can be paused, and the applicant can request a human at any time
Produces a transcript, a score per question, and a recording
Never makes the final hiring decision — that stays with the manager

What the applicant experiences

This is the part that matters most. If the experience is bad, the whole thing falls apart.

A Lily interview goes like this. The applicant applies — often at 9pm after their current shift. Within an hour, the phone rings. A voice introduces itself: "Hi, I'm Lily. I'm an AI calling on behalf of [provider] about your application for the care worker role."

The applicant can ask questions. The AI explains the role, the shifts, the pay, the location. It asks whether they're still interested. If yes, it moves into the structured questions. If they'd rather speak to a person, the AI takes a message and queues it for the team.

Typical questions cover availability, right to work, experience, driving, and one or two situational questions — "A resident is upset because their family haven't visited. What do you do?"

The applicant can take as long as they need. They can ask for a question to be repeated. At the end, they get a clear next step — interview booking, or a respectful no with a reason.

91% of applicants rate this experience positively. Most say it's the first time anyone got back to them on the same day.

What employers get

From the employer side, every AI interview produces:

A full transcript
A recording
Scores per question against the rubric
A summary with flags (e.g. "RTW not confirmed, needs follow-up")
A suggested next step

The manager reads the transcript, listens to the recording if needed, and decides who to interview. The AI hasn't hired anyone. It's narrowed the field with evidence.

What AI interviews shouldn't be used for

Some things AI can't and shouldn't do:

Final hiring decisions. Always human-made.
Complex clinical or senior-role interviews. Use for first round at most.
Anything requiring empathy or disclosure. If an applicant is sharing a difficult personal situation, route to a human.
Anything legally risky without an appeals route. UK GDPR requires a route to human review for consequential automated decisions.

At Lily, we use AI interviews for the first screening round. Managers run the second-round interview in person or by video.

Fairness and bias

AI interviews are only as fair as the rubric behind them. Structured interviews — the same questions, scored the same way — are backed by decades of evidence as the fairest kind of interview. AI makes structure easier to enforce consistently.

But AI can embed bias if:

The scoring rubric is trained on past biased hiring decisions
Speech recognition works worse for some accents than others
The questions themselves disadvantage particular groups

This is why every Lily scoring rubric is signed off by the employer, auditable, and can be changed. Speech recognition is tested across UK, Irish, South Asian, West African, and Eastern European accents, which make up much of the UK care workforce.

Legal and compliance position

Under UK GDPR, automated decisions with significant effects require: a lawful basis, candidate notification, and a route to human review. Lily tells every applicant they're speaking to an AI, explains recording, and flags a human route. The final decision is always human.

CQC safer recruitment rules (regulation 19 for care settings, equivalent safeguarding expectations for Ofsted-regulated children's services) focus on completed checks and documented decisions. A well-designed AI interview produces stronger audit evidence than most human processes.

Frequently asked questions

Are AI interviews legal in the UK?

Yes, with the safeguards above. Many large UK employers — in retail, hospitality, healthcare, and care — now use them for first-round screening.

Do candidates prefer AI or human first-round interviews?

The data surprises people. 91% of Lily-screened candidates rate the experience positively. They prefer a same-day call with an AI to a 9-day wait for a human.

Can candidates cheat with AI tools of their own?

Uncommon in care hiring. Most questions are situational or availability-based — hard to fake convincingly over the phone.

What happens if the AI gets something wrong?

There's a human review route on every decision. Transcripts and recordings mean errors are auditable and fixable, which is rarely true of human screening.

Does the AI make the hiring decision?

No. The hiring manager does. The AI narrows the field with structured evidence.

Next step

Book a demo or read How AI phone screening works — and what it means for candidates.

What is an AI interview? How they work, and what applicants should know