Innovation Profs - 12/19/2023

Your weekly guide to generative AI tools and news

Author

Innovation Profs
December 19, 2023

Sign up for Jan. 26 workshop

We’re hosting an Intro to Generative AI Workshop Jan. 26 at Drake University (and virtually). Sign up by Dec. 29 to get the early bird rate and save $100 (that’s 50 percent off the regular rate).

We’re also booking speaking engagements for 2024. If you are hosting a conference or have a group that would benefit from hearing about Generative AI tools, email us at [email protected].

Generative AI News

The biggest generative AI stories of the week include a number of new products announced by Google.

Google debuts Imagen 2 with text and logo generation

If you’ve closely followed the rapidly shifting generative AI landscape, you’re likely aware that Google has been playing catch-up in the area of AI image generation. Select Google Cloud users have now been granted access to Imagen 2, Google’s new AI image generation model. The key differentiator compared to other models? According to Vishy Tirumalasetty, head of generative media products at Google, “Imagen 2 can generate … emblems, lettermarks and abstract logos … [and] has the ability to overlay these logos onto products, clothing, business cards and other surfaces.” It can also render text in different languages, currently Chinese, Hindi, Japanese, Korean, Portuguese, English and Spanish, a feature not seen in other models.

If you like ChatGPT, then you're going to love Spotify's upcoming feature

TikTok user @robdad_ stumbled across a new feature in Spotify (not yet a widely available feature): a generative AI tool that gives users the ability to generate a playlist using a text prompt. The feature was later confirmed to be real by Spotify. The text-to-playlist prompting pipeline appears to be a novel application of generative AI.

Midjourney Alpha is here with AI image generations on the web

Have you wanted access to Midjourney without having to use Discord? That access may be coming soon. Midjourney just launched its web-based Midjourney Alpha, which allows users to create images in a web-interface. The downside: it’s only currently available to Midjourney users who have created more than 10,000 images (that’s a LOT of images!). Midjourney did announce that the company would be making the tool more widely available in the coming month.

Google’s new AI tool ‘MusicFX’ composes music with just a few words

Google has announced MusicFX, a new text-to-music generative AI tool built on top of an earlier tool, MusicLM, and Google DeepMind’s SynthID, which embeds a unique digital watermark in the tool’s outputs (“ensuring the authenticity and origin of the creations”). To protect original artists, users are also prohibited from including specific artists in their prompts.

Google unveils MedLM, a family of healthcare-focused generative AI models

Back in October we reported that Google announced new generative AI-powered search capabilities for doctors. Now Google has announced MedLM, a family of models fined-tuned for use in the medical industry. Two models are currently available, a larger one for more complex tasks and a smaller one that can be fine-tuned for use across a range of tasks. Possible use cases are reported in the article: “Google says that one early MedLM user, the for-profit facility operator HCA Healthcare, has been piloting the models with physicians to help draft patient notes at emergency department hospital sites. Another tester, BenchSci, has built MedLM into its ‘evidence engine’ for identifying, classifying and ranking novel biomarkers.”

Quick Hits

Exclusive: Jaxon AI teams up with IBM watsonx in battle against AI hallucination

Jaxon AI's Domain-Specific AI Language (DSAIL) technology is designed to prevent hallucinations and inaccuracies with IBM watsonx models.

venturebeat.com/ai/exclusive-jaxon-ai-teams-up-with-ibm-watson-in-battle-against-ai-hallucination

With AI Studio, Google launches an easy-to-use tool for developing apps and chatbots based on its Gemini model | TechCrunch

After announcing its family of Gemini models last week and bringing it to its Bard chatbot experience, Google is now bringing Gemini to developers through new services.

techcrunch.com/2023/12/13/with-ai-studio-google-launches-an-easy-to-use-tool-for-developing-apps-and-chatbots-based-on-its-gemini-model

The age of weaponized LLMs is here

LLMs are the new power tool of choice for rogue attackers, cybercrime syndicates, and nation-state attack teams.

venturebeat.com/security/the-age-of-weaponized-llms-is-here

Anthropic leads charge against AI bias and discrimination with new research

Anthropic researchers unveil new techniques to proactively detect AI bias, racism and discrimination by evaluating language models across hypothetical real-world scenarios, promoting AI ethics before deployment.

venturebeat.com/ai/anthropic-leads-charge-against-ai-bias-and-discrimination-with-new-research

Scientists tackle AI bias with polite prodding

The troubling presence of racial bias in AI output may be easier to contain than many thought. Scientists at AI research company Anthropic say a little politeness may just do the trick, at least in some instances.

techxplore.com/news/2023-12-scientists-tackle-ai-bias-polite.html

Should Artists Disclose Their Use of AI? One Artist’s Work Sparked a Web3 Debate

A digital artist may have used AI to create a “digital oil painting” that was sold earlier this week. NFT artists are debating whether she should have disclosed this use of AI rather than attempting to hide it.

nftnow.com/ai/should-artists-disclose-their-use-of-ai-one-artists-work-sparked-a-web3-debate

How AI changes the game for software makers

Zeev Farbman, CEO of Lightricks, a mobile image and video editing tool, explains how generative AI is trivializing some of the work that formerly required specialized software and expertise.

www.axios.com/2023/12/15/ai-software-makers-change-lightricks-adobe

Tool of the week: Microsoft SwiftKey

Now you can use AI image generation in your favorite mobile apps. Download Microsoft’s Swiftkey AI Keyboard, and you’ll have the power of DALL-E 3 built into your mobile keyboard.

  1. Download here (iOS and Android).

  2. Enable SwiftKey in Settings → Keyboards

  3. In an app such as iMessage, click the globe icon to switch to the Bing keyboard. Click the GIF button and then the create icon to access DALL-E 3.

Innovation Profs Homework

We’re on Instagram. Give us a follow @innovationprofs. And feel free to share your AI-generated images with us to share on the app.

AI-generated image of the week

We tested out photos of Santa Claus on four leading image creation tools - Ideogram, Midjourney, DALL-E 3 and Imagine With Meta. Surprisingly, Google’s image tool wouldn’t create a photo of Santa. 

Prompt: photo of Santa Claus, 4k, realistic

Generative AI tip of the week

Midjourney users can use negative keywords to exclude things from your images. Want to create a photo of mountains but without any snow? Try adding (--no snow) to the end of your prompt. You can also adjust the aspect ratio of your images in Midjourney by adding --ar (example: --ar 16:9).

Get starting with Generative AI

New to generative AI? Here are some places to start…

What we found:

X user @umesh_ai made an image with opposing styles in each half. Click the link below to see more examples.