[AI Weekly] Xiao Ai's "Trying Early adopters" AI model
news of the week
Xunfei Spark Model V2.0 Released
On August 15th, iFLYTEK officially released Xunfei Spark Large Model V2.0 , focusing on improving code capabilities and multi-modal capabilities, launching new functions such as code generation, image generation, video generation, and Spark plug-ins, and launching AI programming tools ——Smart programming assistant iFlyCode 1.0.
At the same time, HKUST Xunfei also cooperated with Huawei to launch the " Xunfei Xinghuo All-in-One Machine ", which provides independent and controllable privatization large models based on industry-specific data for customers in key industries such as finance and state-owned enterprises. At present, it has reached pilot intentions with many customers.
According to the official roadmap of HKUST Xunfei, the Xunfei Xunhuo large model will usher in another upgrade on October 24, and the general model will be benchmarked against ChatGPT to achieve Chinese surpassing and English parity.
Xiaomi: Xiao Ai upgrades AI large model capabilities
At Xiaomi Lei Jun’s annual speech on August 14, Lei Jun revealed that the latest 1.3 billion parameter model of Xiaomi’s AI model has been successfully run locally on the mobile phone , and some scenarios can be compared to the results of the 6 billion parameter model running on the cloud.
At present, Xiao Ai has started to upgrade the ability of AI large model and started the invitation test . Xiao Ai, who is supported by a large model, will have better contextual understanding, a new immersive dialogue form, and be able to complete various complex tasks.
The internal beta version of Xiao Ai's large model will be open to users of MIUI 14 and Xiaomi Sound, and the audit internal test will be launched in batches by model.
DingTalk Personal Edition Opens Internal Beta: AI Services Become the Key
On August 16, the personal version of DingTalk officially launched the internal test, and users can apply for the test through the official website.
DingTalk Personal Edition does not have functions such as reading and checking in. The product will be based on AI . Currently, internal test users can experience large-scale model capabilities such as Wenshengwen, Wenshengtu, role-based dialogue, and AI creation for free for a limited time.
In April of this year, DingTalk announced that it will fully access the large model of Tongyi Qianwen, and the personal version will also be tested simultaneously.
Artificial intelligence simulation town AI Town was born
A few days after Stanford's intelligent agent "Smallville" project was open-sourced, the well-known venture capital firm a16z built a simpler, easier-to-deploy, and also open-source artificial intelligence simulation town "AI Town" based on this.
Now, everyone can customize and build their own "Digital Western World" based on the "AI Town" project on Github. In Digital Western World, AI agents will "live" and communicate with each other in the "town" you set , without knowing that they are living in a simulated environment.
If you want to experience it first, you can play the official "virtual town" online to see what the AI agents are doing:
Fun Fact of the Week
AI app that can chat with Jesus and Satan
Recently, a free app called Text With Jesus was released on the App Store, allowing users to chat with "Jesus", "Mary", "Joseph" and others. Users can also pay $2.99 a month to get the chance to chat with "Satan".
The developer said that the AI model used by this APP has been trained on all public versions of the Bible, such as King James Version, New International Version and New American Standard Bible, and aims to help users deepen their understanding of related texts.
A new problem for beauties: how to prove that you are not AI
Japanese Internet celebrity Fujisaki Nagi is often considered "too much like AI" in the photos on SNS, and is collectively asked by netizens to prove that "I am a human being" .
After sharing photos many times but not being recognized as "human", Fujisaki recently appeared on the TV show "Prove Your True Body", showing the "AI" ears and other parts that questioned the hardest-hit areas, calming doubts.
Regarding the AI doubts from netizens, Fujisaki Nagi said with a wry smile: "I wonder if this is in a magical era."