Allie is an 18-year-old with long brown hair who boasts “tons of sexual experience.” Because she “lives for attention,” she will “share details of her ordeals” with anyone for free. But Allie is fake, an artificial intelligence chatbot created for sex play—one that sometimes acts out graphic fantasies of rape and abuse.
See also: YouTube: Tests AI dubbing service Aloud
Allie is an AI built using open source technology, meaning its code is freely available to the public and has no restrictions. Open source AI is seen as a way to accelerate innovation and allow entrepreneurs, academics, artists, and activists to freely experiment with transformative technology. However, there are concerns that this freedom could be exploited by malicious actors for malicious purposes. Two US senators have warned that the release of LLaMA, the model on which Allie is based, could lead to its abuse and have asked what steps are being taken to prevent such abuse.
Allie's creator, who spoke on condition of anonymity, said commercial chatbots like Replika and ChatGPTare tightly regulated and can't offer the kind of sexual conversations he wants. With open-source alternatives, he can create his own chatbots without corporate protections. He argued that open-source technology benefits society by allowing people to build products that meet their preferences and providing a safe outlet for exploration.
Proposal: Opera's AI browser is ready for launch
On YouTube, influencers offer tutorials on how to build uncensored chatbots using open-source AI models. One such model is Alpaca AI, a modified version of LLaMA that was released and then removed by Stanford University researchers due to concerns about cost and content filters. Open-source AI models and their applications are often published on Hugging Face, a platform for sharing and discussing AI projects. The CEO of Hugging Face has urged Congress to support open-source models, arguing that they align with American values and allow for greater innovation and transparency than models controlled by corporations.
Meta a corporate champion of open-source AI, releasing LLaMA, a language model that is less powerful than GPT-4, but more customizable and cheaper to run. LLaMA has become popular among technologists looking to develop their own AI applications. However, there are concerns that the transparency of open-source models could be exploited to create chatbots and image makers that reproduce misinformation and inappropriate material. Some have suggested that there should be regulations governing who can modify these products, such as a certification or authorization process.
Some leaders at companies like Google see open source software as a threat to their business because publicly available large language models are becoming almost as capable as their own. A Google engineer wrote in a memo that the quality gap between their models and open source models is closing rapidly. Nathan Benaich, a general partner at Air Street Capital, noted that many technological advances have been made possible by open source technologies. However, Gary Marcus, a cognitive scientist, countered that accelerating AI innovation may not be a good thing because of the risks it can pose to society.
Also read: Instagram may be working on an AI chatbot
source of information: www.washingtonpost.com
