Good morning, human brains. Welcome back to your daily munch of AI news.

Here’s what’s on the menu today:

• AI successfully performs click-clack-clack-to-text 👂👻 

  Using audio and AI to decipher what you’re typing.

• Cortana bites the dust 💀

  Microsoft kills off Cortana to pave the way for new AI tech.

• AI helps MI5 identify potential terrorists 🎯

  MI5 has developed its own AI tools to analyze online videos watched by suspects.

Sponsor Bot Eat Brain | New here? Subscribe!

APPETIZER

New AI ‘listens’ to keystrokes with 95% accuracy 👂

A British university collab has achieved a breakthrough in training AI: reconstructing text by listening to your keyboard.

The AI can predict typed text with up to 95% accuracy.
Here’s how the researchers pulled it off:

• They used a MacBook Pro to press 36 different keys 25 times each.

• The AI model was able to identify the characters associated with distinct keystroke sounds, since each one had subtle differences in its waveform.

However, the tech isn’t foolproof:

• Changing typing style can notably lower accuracy.

• Software-generated noise or extra keystrokes can also disrupt system accuracy.

Be very, very quiet.

Our take: these types of security exploits seem less than completely reliable compared to a traditional keylogger. Plus, it’s going to be harder to pull off with ambient noise present and a sea of different keyboards that you don’t have training data for. Reminds a bit of this heat-based hack back in 2015 — freaky, but far from the go-to exploit.

BUZZWORD OF THE DAY

MAIN COURSE

Cortana kicks the bucket 💀

Microsoft has decided to 86 its digital assistant app, Cortana, signaling a shift in its AI priorities.

As confirmed on its support page, the company is now channeling its efforts into other AI developments.

No, no more “Hi.”

They plan to focus on Bing Chat, AI features across Windows, and… Edge.

You’ll still see Cortana on features like Outlook and Teams mobile during the transition, so try to be nice to her. Her days are numbered.

Instead of relying solely on Cortana, users are encouraged to explore alternatives, which are all designed to offer centralized AI assistance:

• Windows 11 voice access for hands-free PC control

• the new AI-driven Bing

• Microsoft 365 Copilot

• Windows Copilot

Our take: First it was Clippy, now it’s Cortana’s turn. It was about time, but we’ll remember them both fondly.

A LITTLE SOMETHING EXTRA

AI helps MI5 identify potential terrorists 🎯 

Britain's intelligence agency, MI5, is sussing out potential terrorists by analyzing what they watch online.

These tools help determine whether individuals on the path to violent radicalization might pose a risk, before an attack takes place.

“Bond, we’ve updated your arsenal.”

As part of sniffing out baddies, the new tech detects violence in images, sparing human staff from having to watch disturbing videos (e.g. beheading and torture).

Ken McCallum, the director general of MI5, stressed the importance of math and data science for national security in a recent talk at the University of Glasgow.

Our take: content moderation is the dark underbelly of the internet that no one likes to talk about. Thank goodness, new AI tech can start replacing real humans when it comes to sifting through disturbing content.

On the flip side, we do feel torn about AI watching our every move online… everything has a price.

MEMES FOR DESSERT

Snape in shades is a vibe and a half.

YOUR DAILY MUNCH

Think Pieces

AI will be 'carnage' for wealth management — AI to bring major disruptions to the wealth management industry.

Architects, the job of the past? — apparently, AI is ‘way ahead of humans’ in the architecture field. But will it overtake us?

Will AI ever reach “funny”? — exploring AI’s limits with humor.

Startup News

AI start-up boom — is the bubble set to burst?

AI start-up sparks global insurance scandal — Israeli-based startup Vesttoo’s at the heart of it.

Tenstorrent bags $100M investment from Samsung & Hyundai — AI hardware startup raised $100m in a convertible note funding round, led by the tech giants.

Research

Google now has an AI grammar checker — no more wasting time on typo’d Google searches.

AI could be the early warning detection system for deadly viruses — a system designed to detect possible worrisome variants of SARS-CoV-2 in advance.

GPT-3 was able to match undergrads on reasoning tasks — the study raises questions over whether AI is starting to 'think' like us.

Tools

DeepCode by Snyk — assists developers to write better and clearer code.

Soundful — AI generated, royalty-free music. Perfect for streamers and video creators.

Transformer — autofills what you’re about to say next with your choice of model.

TWEET OF THE DAY

AI is older than most of us remember.

Tag us on Twitter @BotEatBrain for a chance to be featured here tomorrow.

RECOMMENDED READING

If you like Bot Eat Brain, you might like these other newsletters too:

🏥 Healthcare AI News — 5 minutes or less is all it takes to elevate your Healthcare knowledge with this expert-curated weekly AI newsletter. Stay informed and stay ahead.

👨 The Average Joe — Market insights, trends, and analysis to help you become a better investor. We like their easy-to-read articles that cut right to the meaty bits.

💌 Marketing Letter — The newsletter keeping 30k+ marketers in the loop. Read by marketers who work at LinkedIn, Techcrunch, and Disney.

AI ART-SHOW

Until next time 🤖😋🧠