Staying abreast of the rapidly evolving AI industry can be quite a challenge. So, until an AI can take up the task for you, here’s a convenient summary of recent developments in the realm of machine learning, along with noteworthy research and experiments that we haven’t discussed individually.
During its annual Connect conference last week, Meta unveiled a collection of new AI-powered chatbots integrated into its messaging platforms, including WhatsApp, Messenger, and Instagram DMs. Initially accessible to select users in the United States, these chatbots are designed to emulate specific personalities, even mimicking celebrities like Kendall Jenner, Dwyane Wade, MrBeast, Paris Hilton, Charli D’Amelio, and Snoop Dogg.
Meta’s motivation behind these chatbots is to boost engagement on its platforms, particularly among younger users. The introduction of AI-driven personalities reflects a broader trend in the growing popularity of “character-driven” AI.
One example is Character.AI, which offers customizable AI companions with distinct personalities, such as Charli D’Amelio as a dance enthusiast or Chris Paul as a professional golfer. This summer, Character.AI’s mobile app gained over 1.7 million new installs in less than a week, while its web app received over 200 million monthly visits. Character.AI claimed that, as of May, users were spending an average of 29 minutes per visit, surpassing ChatGPT usage by 300% as ChatGPT usage declined.
Character.AI’s rapid rise attracted significant investments, including over $100 million in venture capital from Andreessen Horowitz, valuing the company at $1 billion.
Another noteworthy AI chatbot platform is Replika, which had approximately 2 million users in March, including 250,000 paying subscribers.
Inworld, an AI-driven character success story, is developing a platform to create more dynamic non-player characters (NPCs) in video games and interactive experiences. While Inworld has not disclosed extensive usage metrics, investments from Disney and grants from Fortnite and Unreal Engine developer Epic Games underscore the potential of AI-driven, expressive characters.
Clearly, there is substantial interest in AI-powered chatbots with distinctive personalities. However, the question remains: will AI-driven characters maintain their appeal over time? Meta’s investment in this new bot collection suggests they are banking on it, but the ever-evolving nature of technology introduces uncertainty, and the novelty may eventually fade.
Here are some other noteworthy AI developments from recent days:
- Spotify is testing AI-generated playlists, offering users the ability to create playlists using prompts.
- The compensation for artists contributing to generative AI datasets remains unclear, despite efforts by some vendors like Adobe to establish funds and revenue-sharing agreements.
- Google expands its AI-powered search experience to teenagers and enhances context and false query detection.
- Amazon launches Bedrock for generative AI models and an enterprise tier for CodeWhisperer, its AI-powered code generation service.
- OpenAI is exploring AI hardware projects, and its secondary-market valuation may reach $90 billion with the release of the more powerful GPT-4 model.
- ChatGPT is evolving to include voice and image-based capabilities.
- The recent Writers Guild of America strike highlighted the role of AI in content creation, with new contract provisions addressing AI usage.
- Getty Images introduces a generative AI art tool for safer image generation.
- Adobe launches Photoshop for the web with AI-powered tools like generative fill and expand.
- Amazon invests up to $4 billion in AI startup Anthropic to compete in the AI sector.
As for the potential limits of AI, experts like Anthropic CEO Dario Amodei believe there are currently no insurmountable boundaries, though time will be the ultimate judge.
In the realm of research, a project at the University of Edinburgh explores how neural networks inspired by insect neural systems can help small robots navigate complex environments efficiently. This approach draws inspiration from the remarkable navigation skills of ants and other small insects.
Additionally, Sony aims to enhance skin color recognition systems using a comprehensive color scale and light/dark levels, addressing biases that affect skin tone accuracy in imaging.
Google introduces RealFill, a generative plug-in for image editing that can fill in missing elements based on available information, potentially revolutionizing photo editing.
Lastly, machine learning models are showing promise in accurately characterizing aftershocks following major earthquakes. While they are not yet perfect, these models have the potential to assist seismologists in predicting quakes and aftershocks, emphasizing the importance of preparedness in earthquake-prone areas.
