Microsoft’s new AI grades your speeches and tells you to stop swearing

We’re not sure which kind of stage fright scares us more: choking in front of a live audience filled with human critics, or bombing a big speech in front of an omniscient ‘bot whose native language is all ones and zeroes.

In a move that’s sure to entrench artificial intelligence even further into our lives as one of those low-level yet ever-present background hums that we’ll all end up taking totally for granted (think Apple’s Siri or Amazon’s Alexa), Microsoft is rolling out a new speech-coaching smart feature for its ubiquitous PowerPoint presentation software.

Flip PowerPoint into “rehearsal mode” and Microsoft’s new Presenter Coach will come alive, taking point as the arbiter of whether that big talk you’re sweating bullets over will end up pleasing your robot audience (and, by implication, your human one). Microsoft says its AI critic will judge your cadence, your reliance on fallback words like “um” and “so,” and even gently tsk-tsk you with onscreen prompts if you forget to use inclusive language or revert to form and start dropping F-bombs.

To dial up the stage pressure that one final notch, Presenter Coach will grade your performance once its patient-listener role is over. In addition to real-time “on-screen guidance for pacing, inclusive language, use of profanity, filler words and culturally insensitive phrases,” the AI feedback will spit out “a detailed report with metrics for additional practice.”

The goal, of course, isn’t to inject an extra dose of angst into people who fear public speaking; it’s the very opposite. “Training and feedback are vital to help people gain confidence and improve their public speaking abilities,” Microsoft notes, “and Presenter Coach in PowerPoint has arrived to help you do just that.”

It’s a noble aspiration for sure. But if an AI can tell us how bad we are at something we dread anyway, why doesn’t it follow that thread to its logical conclusion and just do the whole job for us? That might sound far-fetched, but over in the world of art, researchers at Adobe and the University of Maryland have successfully experimented with an AI tool that’ll do just that.

The researchers’ image-drawing LPaintB machine learning system, reports VentureBeat, “can reproduce hand-painted canvases in the style of Leonardo da Vinci, Vincent van Gogh, and Johannes Vermeer in less than a minute,” using an adaptive algorithm that powers “an intelligent painting agent” to artfully recreate — and put its own stylistic spin on — whatever image it’s been shown.

For now, we’ll settle for being speech-coached by a bot, even if we have to get up on stage and say all the lines ourselves. But we’ll be ready to embrace AI with arms wide open when the fateful day arrives when someone unveils the game-changing machine that’ll just do all the talking for us.