Welcome to your

AI Weekly Digest

presented by Louis (What's AI) Bouchard

1️⃣ Guiding Stable Diffusion with your Images (An Image is Worth One Word Explained)


Text-to-Image models like DALLE or stable diffusion are really cool and allow us to generate fantastic pictures with a simple text input. But would it be even cooler to give them a picture of you and ask it to turn it into a painting? Imagine being able to send any picture of an object, person, or even your cat, and ask the model to transform it into another style like turning yourself into a cyborg of into your preferred artistic style or adding it to a new scene.

Basically, how cool would it be to have a version of DALLE we can use to photoshop our pictures instead of having random generations? Having a personalized DALLE, while making it much more simple to control the generation as “an image is worth a thousand words”. It would be like having a DALLE model that is just as personalized and addictive as the TikTok algorithm.

Well, this is what researchers from Tel Aviv University and NVIDIA worked on. They developed an approach for conditioning text-to-image models, like stable diffusion I covered last week, with a few images to represent any object or concept through the words you will send along your images. Transforming the object of your input images into whatever you want! Read more...

Watch the video

2️⃣ [Sponsored] Sign up for Doug Turnbull’s ‘ML Powered Search’ Live cohort on Sphere


Doug leads the entire Search Relevance practice at Shopify. He has spent the last 10+ years writing industry-leading books such as “Relevant Search” (2016) & “AI Powered Search” (2022), co-creating the Elasticsearch Learning to Rank functionality, and helping developing Search teams for startups and Fortune 500 companies. 

Sign up for Doug Turnbull’s exclusive live cohort, starting October 11. Previous cohorts have had students from Apple, Amazon, Spotify, Microsoft, Twitter, Shopify, Glassdoor, and more.

Learn more about ML Powered Search

3️⃣ AI Ethics with Lauren

We saw from DALLE that its generations turned memes that were often focused on a fictional character that would be well represented in the final product, due to an abundance of consistent training data. Now, that could be achieved with your own camera roll!

Using your own images would offer tons of benefits and potential ethical benefits. Since current mainstream social media is so focused on displays of wealth, this would help generate images without expending the actual resources to take the photo. While this use case would only perpetuate this negative aspect of social media, the resources saved from not flying on a private plane to a remote island would be a significant benefit.

On the thread of social media, it is already a challenge that images of people are often retouched and these alterations are not disclosed, leading to a false sense of reality and distorted perception of the self. This is a huge problem for kids and teens who may be more vulnerable to the impressions of these images, harming their personal development. For personalized AI-generated images, a good safeguard would be to require disclosure that these images have been altered to avoid misleading the audience, just as some countries have required of companies’ social media marketing materials. This measure would take the wind out of the sails of the person faking a private plane trip, but reducing misinformation is a worthy cost.

We also have to be careful what fake words we use to label sets of images. Depending on who is choosing this word, personal bias may leak into their one-word descriptions of a set of images and may potentially create misinformation. For example, a friend of mine doesn’t like whales, and she would likely label a whale image set with a negative word when whales are usually positively associated. This might work for her purposes, but would not apply to those of us who love whales (and are right). You can see how this would become dubious when applied to more sensitive topics than sea life!

Overall, this technology is definitely the next step in the evolution of image generation and how we create content. The best way to move forward in this direction is to implement safeguards to avoid the magnification of current infosphere problems.

- AI Ethics segment by Lauren Keegan
Want to get into AI or improve your skills? Click here!
We are already at the end of this AI weekly digest! Thank you for thoroughly going through this iteration! I hope you enjoyed it. Feel free to follow my daily newsletter where I share 2 papers and news with a clear one-liner summary, and our newsletter at Towards AI, sharing the most exciting news, papers, articles, and memes weekly.

If you have suggestions, comments, or other thoughts, you can reach me by replying to this email or directly on Twitter or Linkedin. Don't hesitate to come to chat with more than 20'000 AI enthusiasts on Discord!
I hope the next week wipes away some of your stress and brings new opportunities, challenges and happiness. A happy new week from your friend!

If you would like to support my work financially, you can become a Patreon and receive a cool role in the discord server at the same time!

Share the knowledge and forward this email to a friend using this link:

- Louis Bouchard
Copyright © 2022 What's AI, All rights reserved.

Want to change how you receive these emails?
You can update your preferences or unsubscribe from this list.

Email Marketing Powered by Mailchimp