As 2023 comes to a close, Artificial Intelligence remains front of mind as tech giants continue to develop more powerful language models. The rapid development comes with growing concern for privacy and energy consumption. Meanwhile, arts organizations use technology to stay on top of trends - bringing popular culture into museums or adjusting to changing voice in social media.  And, as the year comes to a close, the Arts Management and Technology Laboratory is taking a break for the next couple of weeks. Check back January 1 for new content, follow us on socials, or rewind to some of your favorite 2023 articles and podcasts.

AI, Art & Pop Culture

At the Musée d’Orsay, a new Vincent van Gogh interactive AI experience called Hello Vincent brings the artist to life, allowing visitors to ask questions about his life and work. The experience was developed by technology group Jumbo Mana as part of an initiative by the museum to make the programs more relevant to the 21st century.

Other museums are making similar considerations, expanding access and interest to masterpieces through pop culture, and van Gogh seems to be at the center of it. At the Van Gogh Museum in Amsterdam, a special 50th anniversary exhibit partnered with Pokémon to highlight the 19th-century Dutch artist’s inspiration from his Japanese contemporaries, which shared a common thread with the inspiration for the popular media franchise. 

 A special edition card was given out at the museum featuring Pikachu in the style of Self-Portrait with a Grey Felt Hat, which was soon discontinued after garnering so much attention that the museum was overwhelmed by traffic.

Earlier this year, the Metropolitan Museum of Art launched the Replica app, in partnership with Roblox. As visitors stroll through the museum, they can collect digital replicas from well-known pieces to dress up their Roblox avatar, including the famous straw hat from van Gogh’s 1887 self portrait. 

The Most Powerful AI Model Yet

Gemini, Google’s most powerful AI tool, launched this month, is said to outpace ChatGPT4 and other competitor models. The “natively multi-modal” AI was trained on a wide array of images, text, video, and audio, meaning it is able to process and respond to a variety of inputs. Improved ability in math, coding, and reasoning have marked Gemini ahead of other AI models on common benchmarks. There are different sizes available, including Nano, an efficient model designed for smartphones, and Pro, a scalable model now integrated with Bard. Gemini Ultra, a large model designed for data centers, is not yet publicly available.

As the AI race continues, Meta has confirmed that Facebook and Instagram utilize photos and text from public social media posts to train AI models, including the Meta AI chatbot released in September. Meta isn’t alone - X, Gmail, Bing, among others all use user data to enhance various AI features. Google has explicitly stated that it will use any publicly available information on the internet to train its AI models, including generative AI. Signing up for some of these services often defaults user permission to a level of data collection, raising concerns around privacy and information security. 

“When they take your information to train AI, companies can ignore your original intent in creating or sharing it in the first place,” a Washington Post article states.

End of Year - Social Media Trends

2023 has marked changes for Meta platforms, including the release of Threads, competing with X, formerly Twitter. This month, Meta announced that Threads would no longer need to be integrated with Instagram, allowing users to delete their Threads accounting, without impacting long-curated Instagram feeds. 

As we look ahead to 2024, Meta shares predictions about coming content trends, and Gen-Z is setting the tone. Key themes include thrifted fashion, fandom (in television, music, or video games), food, self-improvement, activism and more. 

Similarly, Pinterest experienced a great deal of growth this year as Gen-Z users flock to the platform, making up 40% of their active user base. As other platforms become increasingly devise, Pinterest seems to be more positive, creative and aspirational, experts say. As users plan their lives, trips, meals, and vacations on the platform, Pinterest is an early predictor of emerging trends. Their 2024 report forecasts include a resurgence of Jazz, reuse crafting, blue beauty looks, and gothic western design.

AI Energy Use Grows at Alarming Rate

As AI expands, projections for energy consumption rose with it. An October report by Alex de Vries from School of Business and Economics at the Vrije Universiteit Amsterdam estimated that energy consumption from AI data centers would require up to 15 gigawatts of continuous demand, the equivalent of some small countries like the Netherlands or approximately 15 average nuclear power plants. Under current projections, this would increase data center energy usage by 50% by 2027. This may even be an underestimate,  considering the energy required to cool the chips or potential sharp increase in demand. 

Some companies are working on smaller, more energy efficient and ethical models. For example, Microsoft is training a language model on set library texts, rather than the whole of the internet, a move that requires far less power.