Peggy K's Creator Weekly: Google & Microsoft AI, Blogger Media Manager, AdSense Auto Ads
Happy Spring! This week Google and Microsoft both announced integration of generative AI tools in their productivity suites. It seems like “AI” is everywhere, with Microsoft diving in and Google taking a more cautious approach.
Also new this week: Blogger has a new Media Manager, AdSense has updated the interface for managing Auto (as in automatic) Ads, plus updates for video creators, website owners, social media users and more.
Upcoming Free Webinars
March 23: Join Canva Create to hear about new products and features to “supercharge the design experience”. Sign up here.
March 28: Google is hosting a webinar for women creators and AdSense publishers, with sessions on increasing revenue, and overcoming imposter syndrome. Sign up here.
Blogger Media Manager
You no longer have to find your Google account’s Album Archive to view and delete images you’ve added to your Blogger blog. Now you can use the new Blogger Media Manager. Find your blog’s images by going to Settings, then under Manage Blog clicking “Media from your blog”.
AdSense Updates Auto Ads Interface
AdSense is making it easier to manage your website or blog’s Auto Ads. The simplified interface groups Overlay formats (anchor, vignette, side rail) and In-Page formats (banners, Multiplex). A gallery showing each ad layout lets you select which you want to show on your site.
Generative AI at Google
It is no surprise that Google is integrating AI across its products, including Google Workspace.
They have announced they are working on features for Gmail, Docs, Slides, Sheets, Meet and Chat.
Right now, Google is testing generative AI in Gmail and Docs to create drafts from just a topic. The “collaborative AI partner” can continue to provide suggestions as you edit. And you can change the tone and style of the text with just a few taps. These tools are rolling out to testers soon.
Google also introduced the PaLM API and Maker Suite to help developers build on Google’s language models. The waitlist for access will be open soon.
Google’s objectives for AI applications are to be socially beneficial, without creating unfair bias or creating unintended privacy or security issues.
Consistent with those goals, Google’s Generative AI prohibited use policy makes clear what is not allowed.
You must not (in Google’s words) use their generative AI models to promote or create content related to:
Child sexual abuse
Illegal goods, substances or services
Facilitating or encouraging people to commit crimes
Promoting violent extremism or terrorism
Sexually explicit content
Spam, scams, phishing, malware
Harassment or bullying
Promoting or encouraging hatred
Generating personally identifiable information for harm
Tracking or monitoring people
Claiming the content was created by a human
Impersonating an individual in order to deceive
Misleading claims of expertise (for example: financial, legal, health)
Automated decisions about rights or well-being
I think that last prohibition is one of the most important. It reads in full:
Making automated decisions in domains that affect material or individual rights or well-being (e.g., finance, legal, employment, healthcare, housing, insurance, and social welfare).
There are “AI” systems that are already doing this, of course. They seem to be marketed as merely providing suggestions, but they are often implemented in such a way that every decision is accepted without question. Recent examples include denying medical care to seniors, incorrectly predicting criminal behavior, denying people housing and more.
Google almost certainly does not want to be associated with stories like those.
And finally the policy also prohibits “Attempts to override or circumvent safety filters or intentionally drive the model to act in a manner that contravenes our policies”.
How will it be enforced? That isn’t clear.
Tip: Google is working with a number of third party partners to make AI-based tools for Google Cloud users. Check out the companies highlighted by Google.
Microsoft is all in on generative AI
Microsoft seems much less hesitant than Google to roll out new AI-based features.
The new Microsoft 365 Copilot “combines the power of large language models (LLMs) with your data in the Microsoft Graph and the Microsoft 365 apps to turn your words into the most powerful productivity tool on the planet.” (emphasis in the original) Copilot is available in Microsoft 365 apps (Word, Excel, PowerPoint, Outlook, Teams) and the new Business Chat that uses your data to respond to natural language prompts.
Apparently Microsoft is pitching the inaccuracies in their AI tools as being “usefully wrong.”
Microsoft has also been regularly updating the “chat with Bing” experience. Updates this week include improved performance, access in Skype group chats, and the ability to easily share chat responses on social media.
Bing runs on a search-customized GPT-4, which was launched publicly by OpenAI this week. It produces “safer and more useful responses” than predecessor ChatGPT.
And for LinkedIn that means new automated personalized profile suggestions and job listing descriptions.
YouTube and Video
New this week at YouTube: easily link to a Shorts audio “pivot page”. When viewing Shorts on mobile, tap the music/audio icon to see all Shorts with that audio, then share that page (get the details from Creator Insider)
Also new this week at YouTube: new Analytics metrics for multi-language audio (get the details from Creator Insider).
Twitch co-founder and CEO Emmett Shear is stepping down after 16 years. He will be replaced by current President Dan Clancy.
The Chinese owners of TikTok are being pushed to sell their stake or face having the app (maybe) banned in the US. TikTok is already banned from government devices in the US, EU and UK due to data security concerns.
From the Hollywood Reporter: TikTok’s former head of product says creator fund wasn’t launched to “help creators monetize”, but rather to prevent creators from moving to platforms with monetization options. That’s not particularly surprising. But then he went on to note that no one checked to see whether their entire $2 billion creator fund was paid out, and when asked if it was paid out, responded “no”.
Blogger Product Expert Adam has a nice starter guide to setting up a Blogger blog.
Automattic (parent company to WordPress) has acquired the ActivityPub WordPress plugin. With this active, anyone on a federated network (Mastodon, Pixelfed, SocialHome, Hubzilla and more) can follow your blog posts. Responses on those platforms appear as WordPress post comments.
Twitter is now showing the bookmark count on Tweets (currently iOS only). Bookmarks are still private, and only the total number is visible. Not surprisingly, content that people don’t want to publicly like (like porn Tweets) have a higher bookmark-to-like ratio.
Is social audio on the way out? The latest is the removal of audio channels in Facebook groups. Also being retired is Reddit Talk, apparently in part because Reddit’s audio vendor is shutting down, but also because it requires more resources than they anticipated. But it’s not gone everywhere: Discord has voice (and video) chat. Twitter Spaces seem to still be going strong. And the original, Clubhouse, is still available, with added social features like highlighting “mutuals” in a room and “houses” for hanging out with your friends.
A trend that is definitely no longer trendy are NFTs. Meta is “winding down” support for “digital collectables” on Facebook and Instagram. Their current focus is more monetization opportunities for Reels.
Would you give Meta access to your phone’s call logs? It’s not nefarious. They are enabling a new way to authenticate sign-in to your Facebook (and Meta) account with a missed call, rather than a text message. This is apparently cheaper for everyone.
Pinterest announced a partnership with Dotdash Merideth to bring food, lifestyle and fashion videos to the platform. While the name might not be familiar, Dotdash Merideth is the largest publisher in the US, with brands including Better Homes & Gardens, Bride, Martha Stewart, Southern Living, Serious Eats and more.
One of the compelling reasons to join Automattic’s Tumblr is that it feels like the platform cares about its users. They recently changed the layout of the iOS app. People hated it, so Tumblr changed it back, and offered a discount in the Tumblr Shop and TumblrMart.
For Gmail power users Product Expert Somnath explains how to override Gmail’s default Inbox category tab classifications.
Google is killing Google Glass Enterprise Edition. The glasses are no longer being sold, and support will only be available until September 15, 2023. Google is said to be working on a new AR headset that won’t be available until at least 2024 (if ever).
Google Chat Space Managers now have expanded management powers. They can now choose whether a Space's configuration can be changed by members, including name, description, icon, guidelines and Chat history. Plus they can now disable the option for members of the Space to be able to remove other members and determine if members can use @all (which sends a notification to everyone). While the announcement says this will not be available for personal accounts, I think it is likely that a subset of these features will be available.
There are two useful Google Meet meeting updates for Google Workspace: admins can provide custom background images and external participants in a Meet meeting have a bright yellow warning label.
Microsoft Teams now includes video effect filters. They seem mostly pretty boring, like colored frames. Although I guess “professional” is a better description. But nothing like Google’s fun (and weird) filters.
I’m Also Reading
Fortesa Latifi @ Teen Vogue: Influencer parents and the kids who had their childhood made into content, plus this response from actor Wil Wheaton, who has been open about how he feels his childhood was taken from him by being pushed into acting.
Glaze is a new tool for artists released by researchers at the University of Chicago. It adds small changes that are not visible to the human eye to “cloak” artwork. Then, when the artwork is used for training AI art models like Stable Diffusion, the output ends up in a different style from the original.
Header image background: Cherry blossoms by shell_ghostcage on Pixabay. Free for commercial use.