šŸ“ø What is a photo?

PLUS: Deloitte bets big on GenAI

Was this email forwarded to you? Make sure to subscribe for more!

Greetings, fellow humans. šŸ‘‹

This is Not A Bot - the newsletter about AI that was definitely not written by AI. Iā€™m Haroon, CEO of Autoblocks and founder of AI For Anyone, and I share the latest news, tools, and resources from the AI space.

Happy Tuesday. So whatā€™s the deal with that blue couch? And where can I buy it?

Letā€™s dive into itā€¦

Have a product, service, job, event, newsletter, app, book, movie, tool, or anything you'd like to share with over 40k subscribers?Ā 

And since youā€™re here already, check us out on our socials - snacks on us if you also hit follow :)

šŸ§µ In today's edition:

  • šŸ“øĀ What is a photo?

  • āœ–ļø Fake photos = real consequences

  • šŸ’Ŗ Metaā€™s Massively Multilingual Speech

  • šŸ¤‘ AI Fundraising News

(Sponsored)Ā Innovating with AI - Find Your Breakthrough AI Business Idea

Download a free copy of the Find Your Breakthrough AI Business Idea video lesson and cheat sheet. Learn the exact methods we used to launch 2 AI products in 4 weeks.

šŸ¤– Top AI News

šŸ“øĀ What is a photo?

Argus Filch from Harry Potter is spot on.

Photos might have just changedā€¦ forever.

Last week, a research paper cleverly titled DragGan was dropped on Hugging Face. It showed how a simple picture of an animal could be easily edited to face a camera, open its mouth, make it blink, or do anything you wanted.

The interface is intuitive and easy to understand: users can highlight specific parts of an image they wish to alter through a drag-and-drop function, and the generative AI handles any gaps that may arise, preserving ā€œrealism.ā€

Hereā€™s what this means:

  • Photoshop isĀ probablyĀ dead.

  • You donā€™t have to scold your parents when they miss the tip of your finger on the Eiffel Tower.

The project creators plan to open-source this tech and release their code in June.

Curious times ahead.

Read more: Indian Express

āœ–ļø Fake photos, real consequences

The fake photo that went viral on Twitter

The previous story makes me pensive because of this story right here.

A few weeks ago, we spoke about the power of edited images and how they could negatively affect the world. Yesterday, we got to see a glimpse of that.

šŸ‘†ļøĀ This (apparently) AI-generated photo of an explosion near the Pentagon in D.C. went viral on social media, causing the reputable Twitter account @DeItaone to share the news.

The result? Within 4 mins of the news being shared, the stock market dipped 0.26% before bouncing back to normal.

Things didn't get too bad in the grand scheme of things, as the Arlington PD confirmed that the image and accompanying reports were fake relatively quickly.

But in the future, as these technologies get more advanced, it will be much harder to decipher the difference quickly, which could lead to unthinkable amounts of trouble.

All weā€™re saying is - we need PROPER regulation, fast.

Read more: Yahoo News

šŸ’Ŗ Metaā€™s Massively Multilingual Speech

Taken from Metaā€™s Insta channel

Meta is on a roll.

Their latest open-source project, Massively Multilingual Speech (MMS), can recognize over 4,000 spoken languages, with text-to-speech capabilities in over 1,100.

Hereā€™s the gist:

  • Meta used wav2vec 2.0, the companyā€™s ā€œself-supervised speech representation learningā€ model, that can be trained on unlabelled data.

  • They tapped into religious texts like the Bible and other similar texts to help increase the modelā€™s available languages to over 4,000.

Q&A:

  • Does this mean the model produces ā€œreligiousā€ language? No.

  • Most of these religious texts are usually narrated by men. Is that the only voice type available? No. The model performs well with both female and male voices.

Another pretty cool open-source project for the win.

Read more: Meta | Engadget

šŸ¤‘ AI Fundraising News

FlowX.ai raises $35M Series A to allow enterprises to port legacy and newer software into a single system to build and run applications and services.

Kiwi raises $80M to help consumers build credit history and improve financial literacy.

Axelera AI raises $50M Series A to provide improved computer vision applications that are more accessible and user-friendly.

šŸ—žļø Byte size: AI article summaries

Disclaimer: AI is (partially) used to summarize these articles.

Deloitteā€™s Big Bets On Google Generative AI, ā€˜Proactiveā€™ Workforce Transformation [CRN] - Deloitte is increasing its partnership with Google Cloud by working on generative AI solutions to expand opportunities for businesses worldwide. The two companies are collaborating to help clients bring real-world use cases for generative AI, such as creating new customer experiences and helping consumer product companies rethink the way they do marketing. (Read more)

AI will play a role in Wisconsinā€™s 2024 election season [Isthmus] - AI will play a role in Wisconsin's 2024 election season, but it is unclear as to what that role will be. While the Republican Party of Wisconsin has used AI to ethically advertise on social media, Democratic Party Chairman Ben Wikler says he is closely following AI research to understand its role in politics. Pollster Charles Franklin is concerned about the rise of "deep fakes" in political campaigns, as AI-generated news stories and social media posts could overwhelm traditional reporting and erode public trust in the media. (Read more)

šŸ¦ļøĀ Tweet of the day

šŸ¤” Hmmā€¦ what do we make of this?

What did you think of today's newsletter?

Login or Subscribe to participate in polls.

And that does it for today's issue.

As always, thanks for reading. Have a great day, and see you next time! āœŒļø

ā€” Haroon: (definitely) Not A Robot and @haroonchoudery on Twitter

Join the conversation

or to participate.