You pause halfway through an answer to collect your thoughts. The AI starts replying before you're done, so you have to explain that you were still talking.

That sort of interruption can make a conversation tiring, even when the answers are good.

Tavus is working on those small moments with Griffin. In its one-minute video-call study, 26 of 54 participants believed their AI partner was human. Study

The result caught my attention. I like the possibility of talking naturally without having to organise every thought into a perfect prompt.

I am Alex. Welcome to ShortCu8 by Innov8.

Lets Dive Deep🐰🐰.

Judge a conversational AI by how much effort it takes to talk to it.

For a practice session, notice whether you can finish a thought, correct yourself and continue comfortably. A convincing face alone doesn't tell you whether the conversation is useful.

Tavus says Griffin can keep listening while speaking, react to what it sees and adjust when someone interrupts. It calls this approach “human computing”. Launch

I can see the appeal in something as ordinary as preparing for an interview. Imagine practising an answer, realising halfway through that you've started with the wrong example, and beginning again. You'd want your practice partner to give you that space.

I'd also want it to follow the answer well enough to ask a useful next question. Looking interested would only get it so far.

That's how I'd judge this kind of tool: did the session help me improve, and could I concentrate on what I was saying? Whether it fooled me would be a separate question.

Microsoft announced new voice models and streaming transcription on October 1. Its transcription system can produce text while someone is still speaking, and its new voice models support expressive speech across languages. Microsoft

These are separate developments from Griffin. Together, they suggest that companies are spending more effort on the flow of conversation, as well as the answer itself.

For an AI tutor or a practice partner, that could make a difference. You could spend more attention on the subject and less on working out when the system expects you to speak.

I find that possibility more useful than another demo of a perfectly polished person talking at the camera. I want to know how well a system handles a conversation when I lose my place or change my mind.

Tavus recruited participants who expected a one-minute call with another participant. They were asked afterward whether their partner might have been AI, then told that it was. The 26-of-54 result came from Griffin-Lite. Methodology

That expectation matters. Someone entering a casual conversation may assess it differently from someone deliberately looking for signs of AI. A minute also leaves plenty untested about longer, more demanding calls.

Tavus describes the result as passing a video Turing test. I'd read that as its description of this particular study, rather than a verdict that AI video calling works convincingly for everyone.

Griffin-Lite remains a research preview for selected testers, with broader release still ahead. Access details

So “kinda solved” captures the excitement. I'd still want to try it through a messy conversation before deciding how useful it is.

I would be interested in an AI practice partner that makes it easier to speak naturally. I'd also want it to identify itself clearly before the call starts.

Once I know I'm talking to AI, I can decide whether I want that experience. Then we can get on with the conversation.

We read every reply - just reply to this email and let us know how we can improve !

#AD1

Your AI is only as good as what you feed it. Feed it a transcript with your product names, acronyms, and numbers spelled wrong, and you spend the afternoon hand-fixing the output.

Wispr Flow Notetaker uses your dictionary and calendar, so product names, acronyms, numbers, and uncommon names come out spelled right. It pops up when your meeting starts and captures Zoom, Google Meet, Teams, and Slack huddles with one click, no bot joining the call.

Then your meetings show up inside Claude or ChatGPT through the built-in connector. No copy paste. Try Notetaker free on Mac and Windows.

Try Notetaker Free

#AD2

The International Quant Championship (IQC) is one of the world's largest quantitative research competitions, bringing together participants from across the globe to test ideas, develop skills and engage with quantitative thinking.

Discover IQC and join a global community exploring quantitative research.

Learn more about WorldQuant BRAIN®