• Posts
  • Why AI Progress Is Unlikely to Slow Down

Why AI Progress Is Unlikely to Slow Down

Also: Alibaba rolls out open-sourced AI model to take on Meta's Llama 2


AI continues to make rapid strides with significant developments in compute power, data quantity and algorithmic efficiency. This advancement is mirrored in the realm of video where AI-driven video editing software are making impactful strides. Alibaba, meanwhile, has launched its open-sourced AI models, Qwen-7B and Qwen-7B-Chat, setting a new competition in the AI field. AI has also made its mark in the semiconductor industry, aiding in diverse applications and end uses. On the tools front, Datadog has unveiled Bits, an AI assistant for quicker application issue resolution, while Uber is creating its own AI chatbot. In venture capital updates, growth intelligence platform Haus secured a $17M Series A funding round to further develop its analytical tools.


  • 📈 Why AI Progress Is Unlikely to Slow Down

  • 🎥 The future of AI is video, and it’s coming at us fast

  • 🦙 Alibaba rolls out open-sourced AI model to take on Meta's Llama 2

  • 🦾 Semiconductor Industry Is Pulling AI Across A Diversity Of End Uses And Applications

The swift advancement of AI is attributed to significant developments in three key areas: compute, data, and algorithms. Compute, referring to the processing power used for training AI systems, has seen vast growth, with models such as OpenAI's GPT-4 requiring an estimated 21 septillion operations. Such immense compute has been made possible due to Moore's law and increased spending on large-scale model training. Furthermore, vast amounts of data, scraped primarily from the internet, have been feeding AI systems, enabling them to build more accurate models. Additionally, algorithmic advancements allow these systems to use computational power more efficiently, making the most out of available compute and data. While concerns exist about potential data limitations in the future, many experts, including Jaime Sevilla, director of Epoch, anticipate the rapid pace of AI progress to continue. The growing capabilities of AI systems also raise concerns around misuse in sensitive fields such as cybersecurity, nuclear technology, and biology.

The future of AI in the realm of video is fast approaching, with significant implications both exciting and concerning. Advanced video editing software like those offered by Lightricks is rapidly evolving, utilizing AI to copy styles across different videos. This is a progression from the company's photo-editing technology, implying a technology continuum. However, this powerful tool, akin to dynamite, has the potential for misuse and currently lacks regulatory oversight. The 2024 elections pose a particular threat with the rise of 'deepfake' videos and misinformation. Zeev Farbman, co-founder and CEO of Lightricks, insists the image creation realm is a mostly solved problem, yet there's still room for improvement in user interface and refining techniques. While consumer AI video has made significant strides, consistency remains a challenge, marking the difference between true reality and artificial representations.

Alibaba Group's cloud computing unit has released two open-source AI models, Qwen-7B and Qwen-7B-Chat, in a bid to compete with Meta Platform's Llama 2 model. These large language models (LLMs), each with 7 billion parameters, are the first to be open-sourced by a major Chinese tech company and are seen as a potential challenge to the market dominance of OpenAI and Google. Aimed at aiding small and medium businesses to incorporate AI, the two models' code, model weights, and documentation will be freely accessible worldwide, although companies with more than 100 million monthly active users will need a license to use them. Alibaba also has versions of these models with a higher number of parameters which have not been open-sourced. This move underpins China's broader efforts to rival US companies in the AI field.

The semiconductor industry is increasingly employing AI across a broad spectrum of applications and end uses, as highlighted during a panel discussion at SEMICON West and the Design Automation Conference. This diverse application of AI is seen within Renesas, where it's viewed as a game-changer, particularly for edge computing. Renesas contributes to this trend through its range of compute devices and supportive features like memory interfaces and high efficiency power. It's noted that AI success relies heavily on customer understanding of their data sets, with Renesas offering a range of support depending on customer capability. While adoption of AI in this industry is growing rapidly, the panel noted the need for making AI easier to adopt. AI's role in improving design complexity management, enhancing productivity, and its use in testing AI-based chips were also discussed, demonstrating the pervasiveness of AI in this industry.

🛠️ AI tools updates

Datadog, the cloud applications monitoring and security platform, has unveiled Bits, an AI assistant designed to facilitate quicker resolution of application issues for engineers. Bits utilizes AI to analyze customer observability data, among other sources, to provide rapid answers, recommendations, and automated remediation steps in an understandable, conversational language. It offers a wide range of functionalities, from answering natural language questions and debugging code-level issues to suggesting code fixes and automatically constructing unit tests. The new AI tool, developed using OpenAI's technology, assists in managing cloud-scale infrastructure by synthesizing vast amounts of data, aiming to increase developer efficiency, expedite incident resolution, and optimize cloud spend. Currently, Bits AI is available in beta.

Uber, the ride-sharing giant, is developing its own AI chatbot, according to CEO Dara Khosrowshahi, marking the company's foray into generative AI technologies. This announcement comes in the wake of Uber's first-ever quarterly operating profit, as revealed in their Q2 earnings report, despite a shortfall in revenue expectations. While details about the chatbot project remain undisclosed, the initiative signifies Uber's ongoing commitment to harness AI and machine learning, previously seen in their ride-matching algorithms. The introduction of an AI chatbot could potentially enhance user experience and contribute to profit growth, provided it brings genuine value to users and addresses their app-based needs.

💵 Venture Capital updates

Haus, an innovative growth intelligence platform combining causal inference with experimentation and AI, has successfully raised $17 million in a Series A funding round led by Insight Partners. This investment, with contributions from Baseline Ventures, Haystack Ventures, Upside Partnership, Octave Ventures and Mantis Venture Capital, will facilitate the growth of the Haus team and further the development of its advanced analytical tools. These tools aim to measure and boost business growth in a world where third-party cookies are becoming increasingly obsolete due to data privacy initiatives. Founded by former Google analytics expert Zach Epstein, Haus provides brands with cutting-edge, cost-effective solutions for business impact measurement. Using first-party data only, brands can conduct experiments on demand to answer crucial business questions. This latest funding will further cement Haus' role as a trailblazer in growth analytics.

🫡 Meme of the day

⭐️ Generative AI image of the day