Connect with us

News

Meta’s Llama 3 AI Model: A Giant Leap in Multilingual and Mathematical Capabilities

Meta’s Llama 3 AI Model, with its multilingual and mathematical capabilities, is challenging the status quo in the AI landscape.

Published

on

Meta's Llama 3 AI Model

TL;DR:

  • Meta’s new Llama 3 AI model boasts 405 billion parameters, enhancing multilingual skills and mathematical prowess.
  • Llama 3 outperforms previous versions and rivals in coding, maths, and multilingual conversation.
  • Meta aims to surpass competitors with future Llama models by 2024.

The Arrival of Llama 3: Meta’s Largest AI Model Yet

On Tuesday, Meta Platforms unveiled its most substantial Llama 3 artificial intelligence model, showcasing impressive multilingual abilities and overall performance that challenges paid models from competitors like OpenAI. The new Llama 3 model can communicate in eight languages, write superior-quality computer code, and solve complex maths problems, as stated in Meta’s blog posts and research paper.

A Giant Among AI Models

With 405 billion parameters, Llama 3 surpasses its predecessor released last year. Although it’s still smaller than leading models offered by competitors, Meta’s CEO, Mark Zuckerberg, is confident that future Llama models will surpass proprietary competitors by next year. The Meta AI chatbot, powered by these models, is projected to become the most popular AI assistant by the end of 2023, with hundreds of millions of users already.

The Race for AI Supremacy

As tech companies compete to demonstrate the capabilities of their large language models, Meta’s Llama 3 aims to deliver significant gains in areas like advanced reasoning. Despite concerns about the limits of such models, Meta continues to innovate and invest in AI technology.

Multilingual and Multitalented

In addition to the flagship 405 billion parameter model, Meta is also releasing updated versions of its lighter-weight 8 billion and 70 billion parameter Llama 3 models. All three models are multilingual and can handle larger user requests via an expanded “context window.” This improvement, according to Meta’s head of generative AI, Ahmad Al-Dahle, will enhance the experience of generating computer code.

AI-Generated Data for Improved Performance

Al-Dahle also revealed that his team improved the Llama 3 model’s performance on tasks such as solving maths problems by using AI to generate some of the data on which they were trained. This innovative approach could pave the way for future advancements in AI technology.

Advertisement

Meta’s Strategic Move

Meta releases its Llama models largely free-of-charge for use by developers. This strategy, Zuckerberg believes, will lead to innovative products, less dependence on competitors, and increased engagement on Meta’s core social networks. Despite some investors’ concerns about the costs, Meta’s commitment to AI development remains steadfast.

Llama 3 vs. The Competition

Although measuring progress in AI development is challenging, test results provided by Meta suggest that its largest Llama 3 model is nearly matching and, in some cases, outperforming Anthropic’s Claude 3.5 Sonnet and OpenAI’s GPT-4o. This competitive edge could make Meta’s free models more appealing to developers.

The Future of Llama 3: Multimodal Capabilities

In their paper, Meta researchers hinted at upcoming “multimodal” versions of the models due later this year. These versions will incorporate image, video, and speech capabilities, potentially rivalling other multimodal models such as Google’s Gemini 1.5 and Anthropic’s Claude 3.5 Sonnet.

To learn more about Meta’s Llama 3 tap here.

Comment and Share

What do you think about Meta’s Llama 3 AI model and its potential impact on the AI landscape? Share your thoughts in the comments below and don’t forget to subscribe for updates on AI and AGI developments.

Advertisement

You may also like:


Discover more from AIinASIA

Subscribe to get the latest posts sent to your email.

Continue Reading
Advertisement

News

The AI Revolution: Asia’s Role in a Transforming World

Explore the AI revolution in Asia, its challenges, and future prospects with insights from Eric Schmidt, highlighting the power of large language models and the importance of speed and risk-taking.

Published

on

AI Revolution Asia

TL;DR:

  • Eric Schmidt predicts AI’s impact over the next two years will be profound, surpassing social media’s influence.
  • Frontier models like OpenAI and Anthropic are leading the AI race, but funding and power are significant challenges.
  • Schmidt suggests a shift from arbitrary language to digital commands will revolutionise programming.

The AI Revolution is Here

Artificial Intelligence (AI) is set to transform the world in ways we can barely imagine. Eric Schmidt, former CEO and Chairman of Google, believes the changes brought by large language models (LLMs) like ChatGPT will be more profound than the impact of social media. This transformation will happen rapidly, within the next two years, and will affect every aspect of our lives.

The Power of Large Language Models

Large language models have reached a stage where they can interpret complex prompts and perform tasks that were once thought impossible. Schmidt highlights three key areas where AI will have a significant impact:

  • Context Windows as Short-Term Memory: Developers can use context windows to give LLMs short-term memory, allowing them to process vast amounts of information quickly.
  • AI Agents: These agents can read, understand, and apply complex concepts, such as chemistry, and use this knowledge to perform tasks.
  • Text to Action: The ability to convert text into actions will revolutionise programming, making it accessible to everyone.

The TikTok Example

To illustrate the power of text to action, Schmidt uses the example of TikTok. If TikTok were banned, he proposes using an LLM to create a copy of the app, complete with users, music, and preferences, all within 30 seconds. This demonstrates the potential of AI to disrupt traditional programming and business models.

Money and Power: The Challenges Ahead

While the potential of AI is immense, there are significant challenges to overcome. Schmidt highlights two main obstacles:

  • Funding: The leading AI companies, such as OpenAI, require vast amounts of money to continue their work. Sam Altman, CEO of OpenAI, estimates the company needs $300 billion.
  • Power: The energy required to power these AI models is beyond the current capabilities of the US power grid. Schmidt suggests partnering with Canada, which has abundant hydropower, to meet these energy needs.

The Race for AI Supremacy

Schmidt notes that the gap between the leading AI models and newcomers is widening. Six months ago, he believed the gap was closing, but now it appears to be growing. This highlights the importance of speed and risk-taking in the AI race. Companies that can move quickly and take calculated risks will have a significant advantage.

The Importance of Speed and Risk-Taking

Schmidt emphasises the need for speed in decision-making during this period of rapid change. He attributes Google’s underwhelming performance in AI to its culture, specifically its work-from-home culture, which he believes slows down decision-making. He also highlights the importance of risk-taking, citing Microsoft’s deal with OpenAI as an example of a risk that paid off.

Asia’s Role in the AI Revolution

Asia is poised to play a significant role in the AI revolution. With its vast population, technological prowess, and innovative spirit, Asia has the potential to lead the way in AI development and implementation. Countries like China, Japan, and South Korea are already investing heavily in AI, and their contributions will shape the future of this transformative technology.

Advertisement

The Future of AI in Asia

The future of AI in Asia is bright. As AI continues to evolve, it will create new opportunities and challenges. Asia’s tech-savvy youth are well-positioned to take advantage of these opportunities and drive innovation in the AI space.

Comment and Share:

What do you think the future of AI holds for Asia? Share your thoughts and experiences with AI and AGI technologies in the comments below. Don’t forget to subscribe for updates on AI and AGI developments.

You may also like:


Discover more from AIinASIA

Subscribe to get the latest posts sent to your email.

Continue Reading

News

AI Showdown: ChatGPT Doubles Users, Meta Hits 400 Million, and Google Reboots Strategy

Explore the rapid growth of AI adoption in Asia, with insights from OpenAI, Meta, and Google. Discover how AI is shaping the future.

Published

on

AI adoption in Asia

TL;DR:

  • ChatGPT usage doubled to 200 million active monthly users.
  • Meta’s AI features reach 400 million monthly users.
  • Google reintroduces AI image generator after addressing controversies.

In the rapidly evolving world of artificial intelligence (AI), the competition among tech giants is heating up. Recent data reveals significant growth in the adoption of AI chatbots and tools, highlighting the increasing importance of this technology in our daily lives. Let’s dive into the latest developments from OpenAI, Meta, and Google.

ChatGPT’s Meteoric Rise

OpenAI, the creator of ChatGPT, has announced that the usage of its iconic chatbot has more than doubled since November, reaching an impressive 200 million active monthly users. This surge in popularity is a testament to the growing interest and acceptance of AI-powered tools among the general public.

  • ChatGPT usage doubled to 200 million active monthly users.
  • 92% of Fortune 500 companies are using OpenAI’s services.

“The numbers could help quiet the small but growing group of naysayers that are asking to see bigger and quicker revenue from AI chatbots and other tools.”

Meta’s AI Expansion

Facebook-owner Meta has also reported a significant increase in the usage of its AI features. With 400 million monthly users and 185 million weekly users, Meta’s AI tools are gaining traction among its vast user base. The company’s Llama model, available for free, has seen a doubling in usage between May and July, highlighting its competitive edge against paid models like ChatGPT and Google’s Gemini.

  • Meta’s AI features reach 400 million monthly users.
  • Llama model usage doubled between May and July.

“The use of AI by Meta’s more than three billion users was ‘growing quickly, and we haven’t even rolled out in UK, Brazil, or EU yet,’ CEO and founder Mark Zuckerberg said in post on Threads.”

Google’s AI Comeback

Google, despite its early lead in developing generative AI, has faced criticism for lagging behind its rivals. However, the search engine giant is making a comeback with its Gemini chatbot, now integrated across all its products. Google has also reintroduced its AI image generator to premium and business customers after addressing previous controversies.

  • Google’s Gemini chatbot is now included across all its products.
  • AI image generator reintroduced after addressing controversies.

“Google earlier this year suspended generating images of people after Gemini was discovered to be creating diverse but historically inaccurate images, such as Asian Nazis during World War II or a George Washington who was Black.”

The Future of AI in Asia

The rapid growth of AI adoption in Asia is a testament to the region’s tech-savvy population and its eagerness to embrace emerging technologies. As AI continues to evolve, it is crucial for companies to stay ahead of the curve and address the ethical and practical challenges that come with this powerful technology. As AI continues to evolve, it is clear that its impact on our lives will only grow stronger. Stay informed, stay curious, and embrace the future of technology.

Advertisement

Comment and Share:

We’d love to hear your thoughts on the future of AI and AGI in Asia. What excites you the most about this technology? Have you had any personal experiences with AI tools like ChatGPT or Meta’s AI features? Share your stories and insights in the comments below. Don’t forget to subscribe for updates on AI and AGI developments.

You may also like:


Discover more from AIinASIA

Subscribe to get the latest posts sent to your email.

Continue Reading

News

The Future of AI: A Landmark Treaty Signed by US, Britain, and EU

The AI Convention, the first international AI treaty, addresses AI’s human rights aspects but faces criticisms for lacking enforceability.

Published

on

AI Convention

TL;DR:

  • The first legally binding international AI treaty will be signed by the US, Britain, EU, and other countries.
  • The AI Convention focuses on protecting human rights from potential AI risks.
  • Critics argue the treaty has been watered down and lacks enforceability.

A New Era in AI Governance

Artificial Intelligence (AI) is changing the world rapidly. From self-driving cars to predictive analytics, AI is everywhere. However, with great power comes great responsibility. That’s why the first legally binding international AI treaty is a big deal. This treaty, known as the AI Convention, will be signed by the US, Britain, EU, and other countries. Let’s dive into what this means for the future of AI.

What is the AI Convention?

The AI Convention is a treaty that focuses on the human rights aspects of AI. It was negotiated by 57 countries and adopted in May. The Council of Europe, an international organisation safeguarding human rights, led this initiative. The treaty aims to address the risks AI may pose while promoting responsible innovation.

“This Convention is a major step to ensuring that these new technologies can be harnessed without eroding our oldest values, like human rights and the rule of law,” said Britain’s justice minister, Shabana Mahmood.

How Does the AI Convention Work?

The AI Convention allows signatories to adopt or maintain measures to give effect to its provisions. This means countries can create their own laws based on the treaty’s principles. However, the treaty has been criticised for being too broad and lacking enforceability.

Francesca Fanucci, a legal expert at ECNL who contributed to the treaty’s drafting, highlighted some flaws:

“The formulation of principles and obligations in this convention is so overbroad and fraught with caveats that it raises serious questions about their legal certainty and effective enforceability,” she said.

Criticisms of the AI Convention

Critics argue that the treaty has been watered down. Fanucci pointed out exemptions for AI systems used for national security purposes. She also noted limited scrutiny of private companies compared to the public sector, calling it a “double standard”.

Advertisement

Despite these criticisms, the AI Convention is a significant step forward in AI governance. It shows that countries are taking AI risks seriously and are willing to cooperate internationally.

The AI Convention vs. EU AI Act

It’s important to note that the AI Convention is separate from the EU AI Act. The EU AI Act is a comprehensive regulation on the development, deployment, and use of AI systems within the EU internal market. The AI Convention, on the other hand, is a broader treaty focusing on human rights.

What’s Next for AI Governance?

The AI Convention is just the beginning. As AI continues to evolve, so will the laws and treaties governing it. Countries will need to work together to ensure that AI is used responsibly and ethically.

AI in Asia: A Growing Landscape

While the AI Convention is a global initiative, Asia is also making strides in AI governance. Countries like China, Japan, and South Korea are investing heavily in AI. They are also developing their own AI regulations and ethical guidelines.

Comment and Share:

What do you think about the AI Convention? Do you agree with the critics, or do you think it’s a step in the right direction? Share your thoughts in the comments below. And don’t forget to subscribe for updates on AI and AGI developments.

Advertisement

You may also like:


Discover more from AIinASIA

Subscribe to get the latest posts sent to your email.

Continue Reading

Trending

Discover more from AIinASIA

Subscribe now to keep reading and get access to the full archive.

Continue reading