Innovation

Breaking boundaries: How gen AI transforms canvases, cookbooks and comedy

March 7, 2024 | By Chris Mullen

about In Tech

In Tech is our regular feature highlighting what people are talking about in the world of technology — everything from crypto and NFTs to smart cities and cybersecurity. 

When we teach artificial intelligence to both understand knowledge and creativity, the words of Albert Einstein resonate loudly: “Imagination is more important than knowledge.” And as humans apply generative AI into our civilization, we’ll continue to broaden our horizons into new frontiers: weird, helpful, frightening and enlightening.

The doors this technology opens are limited only to the combined imagination of gen AI and humans. And we’re seeing that imagination play out in real-time already, with new concepts promising to use AI to change the ways we work, play or even feel.

AI is now reshaping traditional applications but also quirkier and obscure realms as well. We looked at some of those unexpected uses of AI during the past year and explain where the tech can possibly take us.

AI ‘mastering’ the dungeon

With the decline of the stigmas associated with Dungeons and Dragons, the game, once relegated to dank basements in the ‘80s, has exploded in popularity. Not only is it a major plot element on the Netflix series “Stranger Things,”  but the role-playing game is a focus of the popular online series “Critical Role” and the recent 2023 Game of the Year “Baldur’s Gate 3” is set within the D&D universe.

But can AI enhance the experience for the game captains, known as dungeon masters? One Reddit user playing D&D with ChatGPT says it was amazing.

From interpreting game situations to tracking historical statistics to setting up storytelling plot points, the responsibilities of a dungeon master certainly overlaps with generative AI’s ability to create countless engaging scenarios for players.

Many are working to train AI to run games of Dungeons & Dragons, but, as with most applications of this technology, we’re not quite ready to replace the human dungeon master just yet. But there are still tools aplenty to help guide the game. For example, AI art generators from Canva and Midjourney can help visualize scenarios and characters on the fly and in a way we’ve never seen.

Getting ‘critical’ with art

Teaching AI what we like and don’t like in art is an interesting concept. Since paintings, music, sculptures and films are all open to interpretations and appreciation, is it possible for AI to be an art critic? How can this work if art is subjective and how can the AI determine any real opinions about the work itself?  Projects like Critbot have a compelling premise.

Enter AI Art Critics…

“Critbot is trained on the language of numerous human art critics and analyzes submitted artwork by drawing on the knowledge of its extensive database,” its website says. In other words, the AI interprets elements from a submitted image, organizes that metadata and then cross-references data points against the collective art critic knowledge the AI has been fed.

But why even do this? For starters, tools like this can give artists or armchair enthusiasts a basic understanding of how historical critical analysis might view a piece of art. By doing so, users can understand how a future work might be interpreted by others.

Bringing soul into food

OK, so there’s no way to literally feed generative AI food to analyze.  However, there’s no shortage of recipes around the web. Simple recipes for common cuisine can be difficult to find without the unnecessary flair of a blogger, celebrity chef or influencer.  Also, parsing out the recipe from the abundance of backstory, unnecessary information and many adverts can have amateur chefs scratching their heads.

Enter AI recipes. It’s not “Soylent Green,” we promise.  Applications like DishGen have trained their generative AI on the cornucopia of recipes around the web to simplify the experience of obtaining a recipe.

These can range from wildly simple prompts (“recipe for baked beans”) to dietary limited alternatives (“recipe for gluten-free pop tarts”) to even imaginative or complex prompts (“Thai-inspired ravioli”). 

While an AI will presumably never “tastes” any of these dishes, it’s safe to assume that with prolonged training, recipe generation will continue to improve in both culinary complexity and simplicity for the lay chef. A note at the bottom of each AI response implores cooks to leverage common sense above the recommendations of the AI…which seems healthy.

AI hits the funny bone

Here's a bold statement: In its current state, generative AI is terrible at comedy. Sure, it can excel at generating structured “knock-knock” jokes because these simple witticisms have a consistent pattern and structure.

The problem AI has with comedy is that the experience involves a feedback loop from an audience. Audiences react and the comedian learns from this feedback. So, while gen AI can learn simple jokes and even be trained, it can lack the je ne sais quoi that makes the nuance of comedy effective.

Enter the AI comedian. In January, an interesting use case of this technology went viral when a gen AI tool analyzed the work of George Carlin and then wrote a 60-minute concert featuring the deceased comedian using the AI generated script and a voice AI synthesizer.  

The reactions to using AI in this manner were as polarizing as comedy itself, both invoking a protectionist response and defending the memory and dignity of the legendary comic. It also conjured limitless possibilities. 

If anything, we’re at the intersection of technology and imagination. These quirky examples where machines have become dungeon masters, culinary maestros, art critics and even comedians not only showcase the prowess of generative AI, but also invite us to contemplate the profound possibilities that lie ahead. Even if they’re weird, helpful, frightening or enlightening.

Chris Mullen, contributor