•
August 24, 2023
•
6 mins
Hey Niuralogists!
We hope your week is going well! As always, we're back to bring you the most important and exciting updates from the fast-paced world of AI. It's incredible how quickly things are moving, but we're here to help you stay in the loop. In this week's issue, China implements new AI regulation, YouTube experiments with AI music, Meta launches a new AI language tool, Spain creates an AI supervision agency, and Chinese tech companies join the AI chatbot race.
China has become the first major country to enact comprehensive regulations governing generative AI, focusing on national security and adherence to socialist values. The 24-article framework has already led to significant changes in the Chinese market, including the removal of over 100 generative US-based AI apps like ChatGPT from Apple's Chinese App Store. Unlike draft regulations in the United States and Europe, which primarily focus on user safety and transparency, China's rules are geared toward controlling AI to protect national interests, although they also address intellectual property and personal information. These measures are set for review to gauge their effectiveness.
As China aims for a balanced approach between innovation and control, Western nations are also shaping their own AI governance strategies. In the United States, President Joe Biden has met with leading IT executives to discuss AI safety and transparency, while the Organization for Economic Cooperation and Development (OECD) is working on international guidelines.
YouTube is teaming up with Universal Music Group and several artists to launch the Music AI Incubator, aimed at exploring the potential of artificial intelligence in the music industry. Universal CEO Lucian Grainge and YouTube CEO Neal Molan both emphasized the initiative's dual goals of safeguarding copyrights while helping artists profit from AI innovations. This partnership comes in response to rising concerns about unauthorized and unlicensed use of artists' work through AI, highlighted by incidents such as a song featuring AI-generated voices of Drake and The Weeknd going viral and getting pulled from streaming platforms.
While some in the music industry have welcomed the technology, others have warned of the potential for a flood of unlicensed content and copyright violations if AI is not properly managed. Financial analysts from Goldman Sachs and Citigroup are optimistic about the application of AI in the music industry, citing its ability to generate new opportunities while protecting intellectual property. YouTube plans to share further details about AI-related technologies, monetization strategies, and policies in the coming months.
Meta has launched Seamless MT, an AI-powered translation engine that is capable of translating languages in both text and speech forms without needing intermediate models. The engine is particularly advanced in that it can translate spoken words directly from one language to another, bypassing the need for a text conversion step. Released under a Creative Commons license, the engine supports speech recognition, speech-to-text, speech-to-speech, text-to-text, and text-to-speech translations for nearly 100 languages in various combinations.
Paco Guzmán, a research scientist manager at Meta, noted that Seamless M4T's ability to switch languages "on the fly" makes it particularly useful for populations that commonly mix languages in daily conversation. The development puts Meta in competition with Google which has also been actively developing language translation products, including AudioPaLM which focuses on preserving intonation and other speech characteristics during translation.
Spain has established the Supervision of Artificial Intelligence (AESIA), marking the first dedicated AI regulatory body in the European Union. Announced through a royal decree, AESIA aims to align with Spain’s National Strategy on Artificial Intelligence (ENIA) and its broader Digital Agenda 2026, both of which seek to create an "inclusive, sustainable, and citizen-centered" AI landscape.
The agency's formation also comes ahead of the European Union's working group focused on AI regulations, following the EU parliament's approval of a new Artificial Intelligence Act draft in June. As part of Spain's Recovery, Transformation, and Resiliency Plan initiated in June 2021, AESIA aims to position Spain as a pioneer in the AI sector. While the impact of AI on the job market remains a subject of debate, Spain's proactive stance signifies its commitment to shaping AI's role in society.
The Chinese tech giants Tencent and Alibaba are accelerating their efforts in the AI sector as they aim to challenge Western companies like Google, OpenAI, and Meta. Tencent announced plans to launch its own foundation AI model, Hunyuan, later this year, with the goal of making it one of the best in China. The model is being tested across various domains including gaming, advertising, cloud computing, and fintech. At the same time, Alibaba has released two open-source AI models, Qwen-7B and Qwen-7B-Chat, positioning them as rivals to Meta's Llama 2 and OpenAI's ChatGPT 3.5. These models will be freely available to developers worldwide, although larger entities with over 100 million users will require a license.
This aggressive push for AI dominance comes as Beijing encourages its tech firms to develop advanced, "controllable" models to compete internationally. The push is further fueled by estimates suggesting that AI could contribute over $15 trillion annually to the global economy by 2030. Despite these strides, Chinese models have not yet reached the scale of their Western counterparts with Alibaba's models not being as large as Meta's Llama 2.
📬 Receive our amazing posts straight to your inbox. Get the latest news, company insights, and Niural updates.
💣How can AI be used in the military?
The growing field of AI has found its way into military applications. This development is displayed by Palantir's recent demonstration of its AI Platform (AIP). Palantir’s AIP is capable of deploying Large Language Models (LLMs) and other AI algorithms across varying network classifications to assist military operators in real-time decision-making. By integrating and classifying data from diverse sources, both public and classified, AIP offers a comprehensive operational picture. It also handles complex queries that could range from identifying enemy units to recommending tactical maneuvers and managing weaponry. Notably, AIP also maintains a secure digital record of all operations, linking AI-generated responses back to their underlying data for full transparency. Although AI technology’s adoption comes with ethical and regulatory concerns, this technology holds the potential to revolutionize military capabilities.
🔐Can AI artwork be copyrighted?
The question of copyrighting AI-generated art was finally answered when a U.S. federal judge, Judge Beryl Howell, ruled that such works are not eligible for copyright protection. This decision emerged from a lawsuit filed by computer scientist Stephen Thaler, who argued for the copyrightability of an image generated by his AI software. The court's rationale is anchored in the notion that a "guiding human hand" is essential for a work to be copyrighted, as stated by the U.S. Copyright Office, which emphasizes the importance of the "nexus between the human mind and creative expression." Although the ruling allows for the copyrighting of "AI-assisted" art, where human involvement in selection or arrangement is significant, it makes clear that purely AI-generated works do not meet the requisite criteria for copyright protection. In conclusion, AI-generated artwork cannot be copyrighted and remains in the public domain within the United States.
🎼How will AI impact the music industry?
The use of AI in the music industry has already made a large impact and continues to grow. Platforms like LANDR and Izotope leverage machine learning to automate mixing and mastering which streamlines a process that once took humans days. Additionally, analytical AI tools are similarly transforming music promotion by sifting through data to uncover marketing opportunities, much like advanced AI algorithms are designed to mimic human-like reasoning. However, there are ethical dilemmas with AI-generated songs involving unauthorized emulation of artists such as the instance AI song using the voices of Drake and The Weeknd. Moreover, AI music composition, as demonstrated by Sony’s AI-generated song "Daddy’s Car," adds another layer of complexity. Despite all the benefits and features, AI lacks emotional depth and authentic human experience, similar to how AI models like LaMDA are criticized for lacking genuine sentience due to an absence of subjectivity and originality.
AI Tools:
