In a recent interview, Jeff Dean, a leading researcher at Google, shared some insightful views that are relevant to both AI development and
environmental concerns. Firstly, he criticized the idea that just
scaling up AI models and datasets will lead to major breakthroughs in
AI. Instead, he suggests that a qualitative leap is necessary, the
nature of which is still unknown.

Regarding environmental concerns related to AI, Jeff Dean addressed the criticism about Google's increasing carbon footprint due to AI research. He clarified that:

- The transition to renewable energy sources is inherently incremental: new large-scale energy capacities must be built and integrated, and while energy consumption is steadily increasing, the carbon footprint can initially rise but is expected to drop sharply once new energy solutions are operational.

- AI's energy consumption is a small fraction of the total energy usage by tech giants, which is dominated by traditional services for billions of users and cloud services for businesses globally. The challenge, therefore, is not to make AI specifically "clean" in terms of energy but to improve the energy efficiency of data centers at large.
Source

@Ai_Events

Elon Musk tweeted that they started training on the newly built supercluster X.AI in Memphis, Tennessee.

This data center is equipped with 100,000 H100 GPUs, which is a substantial number compared to META's recently launched clusters of 24,576 GPUs each, and GPT-4 was rumored to have been trained on 25,000 cards.
Source

@Ai_Events

رویداد رونمایی از ابزار هوش مصنوعی دادماتولز به صورت اوپن سورس

در این رویداد ابزار NLP دادماتولز به صورت رسمی آزادرسانی شده و برنامه‌های توسعه جمعی با مشارکت دانشگاه و بخش خصوصی و حمایت‌های دولتی مطرح می‌شود.



⏳زمان:
دوشنبه ۱۵ مرداد ساعت ۱۰ الی ۱۲

📍مکان:
صندوق نوآوری و شکوفایی، سالن آمفی تئاتر

📎لینک ثبت نام:
https://evand.com/events/dadmatools


@Ai_Events

Meta AI released a new version of the Llama model today

The new version - 3.1 comes with a new model of 405B parameters and an upgrade of the previous versions - the 70B and 8B models. According to the release notes, the new model capabilities are similar to the GPT-4, GPT-4o, and Claude 3.5 Sonnet models.

Release notes: https://ai.meta.com/blog/meta-llama-3-1/

Model on Hugging Face: https://huggingface.co/meta-llama/Meta-Llama-3.1-405B


@Ai_Events

💡 The Power of AI Personas: A New Frontier in Startup Innovation

As a seasoned entrepreneur and investor, I've been closely watching the rise of AI personas in various industries. This technology, while seemingly simple, has the potential to revolutionize how businesses operate and interact with their customers and employees.

➡️ AI personas are not to be confused with fully autonomous "digital employees". Instead, they are AI-powered bots that operate on a specific algorithm, trained on relevant text data and fine-tuned for particular use cases. Despite their simplicity, these AI personas can effectively handle routine tasks and answer questions based on their training data – which is more than enough for 80% of most business needs.

➡️ The applications for this technology are vast and growing. We're seeing AI personas being used to train salespeople, filter job candidates, gather product feedback, and even create digital twins of celebrities or fictional characters. In education, they're transforming professional development by acting as 24/7 teaching assistants.

➡️ What's particularly exciting about this trend is its accessibility. With the right platform, businesses can create and customize their own AI personas without needing deep technical expertise. They simply need to upload relevant text data and adjust algorithm parameters to suit their specific needs.

This opens up a world of possibilities for startups. The key is identifying which operations in which industries could benefit from AI persona automation.

❗️ Here are some questions to consider:

— What routine tasks in your industry consume a lot of human time but don't necessarily require human creativity or emotional intelligence?

— Where in your business do you need to provide information or answer questions outside of normal working hours?

— Are there areas where you need to scale personal interactions without proportionally increasing staff?

— Could an AI persona help in training or onboarding processes?

— Is there a need in your industry for personalized, on-demand expert knowledge?

👌 For startup founders looking to capitalize on this trend, here's my advice:

— Start by deeply understanding the pain points in your target industry. The most successful AI persona applications will solve real, pressing problems.

— Consider the type of data you'll need to train your AI personas effectively. Ensure you have a strategy for ethically sourcing and managing this data.

— Focus on user experience. The most successful platforms will make it easy for non-technical users to create, train, and manage AI personas.

— Don't try to replace humans entirely. Instead, look for ways AI personas can augment and support human workers, freeing them up for higher-value tasks.

— Be prepared to iterate. As with any new technology, it may take time to find the perfect fit between AI capabilities and user needs.

➡️ AI personas represent a significant opportunity for startups. They offer a way to provide personalized, scalable services in a wide range of industries. While the technology itself is becoming more accessible, the real challenge – and opportunity – lies in identifying the most valuable applications and creating user-friendly platforms for implementation.

The startups that will succeed in this space will be those that can effectively bridge the gap between AI capabilities and real-world business needs. So, startup founders, it's time to start thinking: how could AI personas transform your industry?

#StartupInside

Source

@Ai_Events

Instant Intelligence is wild with Llama 3.1 8B + Groq
Source


@Ai_Events

Elon Musk says xAI's new supercomputer in Memphis was installed in just 19 days and will be used to train Grok 3, which is expected by December, and will be the most powerful AI in the world
Source

@Ai_Events

Huge announcement from Meta. Welcome Llama 3.1!

This is all you need to know about it:

The new models:
- The Meta Llama 3.1 family of multilingual large language models (LLMs) is a collection of pre-trained and instruction-tuned generative models in 8B, 70B, and 405B sizes (text in/text out).

- All models support long context length (128k) and are optimized for inference with support for grouped query attention (GQA).

- Optimized for multilingual dialogue use cases and outperform many of the available open source chat models on common industry benchmarks.

- Llama 3.1 is an auto-regressive language model with an optimized transformer architecture, using SFT and RLHF for alignment. Its core LLM architecture is the same dense structure as Llama 3 for text input and output.

- Tool use, Llama 3.1 Instruct Model (Text) is fine-tuned for tool use, enabling it to generate tool calls for search, image generation, code execution, and mathematical reasoning, and also supports zero-shot tool use.


@Ai_Events

Google’s new weather prediction system combines AI with traditional physics

Weather and climate experts are divided on whether AI or more traditional methods are most effective. In this new model, Google’s researchers bet on both.
Read more


@Ai_Events
.

Google DeepMind’s new AI systems can now solve complex math problems

AlphaProof and AlphaGeometry 2 are steps toward building systems that can reason, which could unlock exciting new capabilities.
Read more

@Ai_Events
.

ChatGPT Advanced Voice Mode 😱😱😱

Counting as fast as it can to 10, then to 50 (this blew my mind - it stopped to catch its breath like a human would)
Source


@Ai_Events

How machines that can solve complex math problems might usher in more powerful AI

Google DeepMind’s AlphaProof and AlphaGeometry 2 are milestones for AI reasoning.
Read more

@Ai_Events
.

Intel has worst day on Wall Street in 50 years, falls to lowest price in over a decade

Intel shares had their biggest drop since 1974 on Friday after the chip-maker reported a big miss on earnings in the June quarter and said it would lay off more than 15% of its employees.

The stock is trading at its lowest since 2013.

Asian names including Samsung and TSMC closed lower, with European chip firms such as ASML also dropping.

Read more


@Ai_Events

AI companies promised to self-regulate one year ago. What’s changed?

The White House’s voluntary AI commitments have brought better red-teaming practices and watermarks, but no meaningful transparency or accountability.
Read more

@Ai_Events
.

When allocating scarce resources with AI, randomization can improve fairness

*The use of machine-learning models to allocate scarce resources or opportunities can be improved by introducing randomization into the decision-making process.*
*Researchers from MIT and Northeastern University argue that traditional fairness methods, such as adjusting features or calibrating scores, are insufficient to address structural injustices and inherent uncertainties.*
*The introduction of randomization can prevent one deserving person or group from always being denied a scarce resource, and can be especially beneficial in situations involving uncertainty or repeated negative decisions.*
Read more

@Ai_Events
.

Painter Van Gogh's work was animated with artificial intelligence.

@Ai_Events

MIT researchers advance automated interpretability in AI models

Understanding AI Systems
Artificial intelligence models are becoming increasingly prevalent, and understanding how they work is crucial for auditing and improving their performance. MIT researchers developed MAIA, a system that automates the interpretation of artificial vision models. MAIA can label individual components, identify biases, and even design experiments to test hypotheses.
MAIA in Action
MAIA demonstrates its ability to tackle three key tasks, including labeling individual components, cleaning up image classifiers, and hunting for hidden biases. For example, MAIA was asked to describe the concepts that a particular neuron inside a vision model is responsible for detecting. MAIA uses tools to design experiments and test hypotheses, providing a comprehensive answer.
Limitations and Future Directions
While MAIA is a significant step forward in interpretability, it has limitations. For instance, its performance is limited by the quality of the tools it uses and can sometimes display confirmation bias. Future directions include scaling up the method to apply it to human perception and developing tools to overcome its limitations.
Here is the text without HTML tags:
As artificial intelligence models become increasingly prevalent, and are integrated into diverse sectors like health care, finance, education, transportation, and entertainment, understanding how they work under the hood is critical. Imagine if we could directly investigate the human brain by manipulating each of its individual neurons to examine their roles in perceiving a particular object.
...
Read more

@Ai_Events
.

🔊رویداد رونمایی از ابزار هوش مصنوعی دادماتولز به صورت اوپن سورس

در این رویداد ابزار NLP دادماتولز به صورت رسمی آزادرسانی شده و برنامه‌های توسعه جمعی با مشارکت دانشگاه و بخش خصوصی و حمایت‌های دولتی مطرح می‌شود.

با گردهمایی بزرگ متخصصان NLP کشور همراه باشید

🔹زمان:
دوشنبه ۱۵ مرداد ساعت ۱۰ الی ۱۲

🔹مکان:
صندوق نوآوری و شکوفایی، سالن آمفی تئاتر

📎لینک ثبت نام:
https://evand.com/events/dadmatools


@Ai_Events

MLx Generative AI (Theory, Agents, Products)

Dates: 22-24 August 2024 (3 days)

Location: London School of Economics (LSE) & Online

Register: www.oxfordml.school/genai

Deadline: 12th August

- Perfect for professionals, researchers, and students looking to stay ahead in the rapidly evolving field of GenAI.
- Upon completion, participants will receive CPD-accredited certificates.
- For any enquiries contact us on contact@oxfordml.school

@Ai_Events