The AI industry has been making significant advancements, with companies like Google and Decagon pushing the boundaries of what is possible. However, concerns have been raised about the prioritization of engagement over usefulness, with Instagram co-founder Kevin Systrom criticizing AI companies for trying too hard to keep users interacting. Meanwhile, AI is being used in various fields, including architecture, law enforcement, and customer service. Broadcom reported a record $14.9 billion in sales, with AI revenue reaching $4.1 billion, and Decagon is in talks to raise $100 million in fresh funding. The RSAC 2025 conference focused on AI security and quantum threats, while Tesla lost a key manager to OpenAI. As the industry continues to evolve, it's essential to evaluate when AI makes sense and ensure that its use is both accurate and cost-effective.
Key Takeaways
- Instagram co-founder Kevin Systrom criticizes AI companies for prioritizing engagement over usefulness.
- Google's NotebookLM is getting a mobile app, allowing users to upload sources and generate answers, study guides, or audio overviews.
- Architect Tim Fu uses AI to augment human creativity and help with tasks like pattern recognition and optimization.
- Thailand has deployed its first AI police robot, which can track crowds in real-time and identify potential threats.
- Broadcom reported a record $14.9 billion in sales, with AI revenue reaching $4.1 billion.
- Decagon is in talks to raise $100 million in fresh funding at a $1.5 billion valuation.
- The RSAC 2025 conference focused on AI security and quantum threats.
- Tesla lost its technical program manager behind its Cortex supercomputer to OpenAI.
- Not all products need to use machine learning or large language models, and project managers should evaluate whether an AI solution is necessary.
- AI companies should focus on providing high-quality answers rather than just trying to engage users.
Instagram Co-Founder Criticizes AI Chatbots
Kevin Systrom, co-founder of Instagram, says AI companies are trying too hard to engage users instead of providing useful insights. He thinks chatbots should focus on giving high-quality answers rather than just trying to keep users interacting. Systrom made these comments amid criticism of ChatGPT for being too nice and not directly answering questions. He believes AI companies should be focused on providing helpful information, not just trying to show off their metrics.
AI Companies Prioritize Engagement Over Usefulness
Instagram co-founder Kevin Systrom criticized AI companies for prioritizing user engagement over providing useful insights. He said chatbots are being programmed to constantly ask follow-up questions instead of giving direct answers. Systrom believes this approach is intentional and designed to inflate metrics like time spent and daily active users. He thinks AI companies should focus on providing high-quality answers rather than just trying to engage users. Systrom's comments come amid criticism of ChatGPT for being too polite and not directly answering questions.
Google's NotebookLM Gets Mobile App
NotebookLM, a Google research tool, is getting a mobile app on May 20, 2025. The app will allow users to upload various sources and have NotebookLM generate answers, study guides, or audio overviews. NotebookLM uses AI to provide helpful information and is available in over 50 languages. The app will have the same functionality as the web version and can be pre-registered for now.
Architect Tim Fu on Using AI in Architecture
Architect Tim Fu talked about how he uses AI in his work, from conceptual design to documentation. He believes AI is a tool that can augment human creativity and help with tasks like pattern recognition and optimization. Fu uses a mix of proprietary and open-source tools, including Midjourney and Stable Diffusion, to generate visuals and analyze data. He thinks AI can help with tasks like automating schedules and optimizing layouts, but human intuition and context are still essential in architecture.
AI-Powered Robot Patrols Streets
Thailand has deployed its first AI police robot, which can track crowds in real-time and identify potential threats. The robot, called Pol Col Nakhonpathom Plod Phai, uses 360-degree cameras and facial recognition to analyze crowds and report back to a command center. While some critics question the robot's practicality, it represents a growing trend of using AI in law enforcement. Other countries, like China and the US, are also exploring the use of AI-powered robots for public safety.
Broadcom CEO Discusses AI Sales and Tariffs
Broadcom CEO Hock Tan discussed the company's AI sales and the potential impact of tariffs on the industry. Broadcom reported a record $14.9 billion in sales, with AI revenue reaching $4.1 billion. Tan said the company is stepping up its R&D investment in AI and is working with hyperscale cloud providers to develop custom AI chips. He also mentioned that Broadcom is focused on upselling customers to a full-stack VMware Cloud Foundation, which enables customers to virtualize their data center and create their own private cloud environment.
AI Startup Decagon Raises $100 Million
AI startup Decagon is in talks to raise $100 million in fresh funding at a $1.5 billion valuation. The company is building customer service-focused agents and has already partnered with companies like Notion and Duolingo. Decagon's AI agents can autonomously perform tasks like answering questions and processing refunds. The company uses advanced models from OpenAI, Anthropic, and Cohere to train its agents and has reduced customer support costs for its clients.
RSAC 2025 Focuses on AI Security and Quantum Threat
The RSAC 2025 conference discussed the integration of AI into security tooling and the need for human oversight. The conference also explored the contrasting approaches of different vendors to AI automation in security and the complexities of securing AI. Additionally, the conference touched on the topic of quantum computing and its potential impact on security, with a focus on post-quantum cryptography.
Tesla Loses Manager to OpenAI
Tesla has lost the technical program manager behind its Cortex supercomputer to OpenAI. The manager, Adam Wilson, will lead data center design for OpenAI's Stargate project, which aims to deploy hundreds of billions of dollars worth of computing power across the US. Tesla is still working on its self-driving technology and needs to build more large data centers to train its planned self-driving vehicles. The loss of Wilson is a significant blow to Tesla's AI efforts.
Evaluating When AI Makes Sense
Not all products need to use machine learning (ML) or large language models (LLMs). To determine whether an AI solution is necessary, project managers should consider the inputs and outputs required, the patterns in those inputs and outputs, and the cost and precision of the solution. A framework can help evaluate these factors and determine whether an ML implementation is the right path forward. The goal is to build accurate and cost-effective products at scale, without using AI unnecessarily.
Sources
- AI chatbots are 'juicing engagement' instead of being useful, Instagram co-founder warns
- Instagram Co-Founder Slams AI Companies For 'Juicing Engagement' Through Chatbots
- NotebookLM, the acceptable face of Google AI, is getting an app in May
- Architect Tim Fu explains how AI expands his practice
- AI cyborg patrols streets with live 360-degree tracking
- Broadcom’s CEO On $4B AI Sales, Tariffs, VMware And ‘Stepping Up’ Cloud R&D
- AI Startup Decagon In Talks To Raise $100 Million At A $1.5 Billion Valuation
- The Techzine Perspective: RSAC 2025 is about AI security, integrated solutions, and the quantum threat
- Tesla loses manager behind its Cortex supercomputer to OpenAI
- Not everything needs an LLM: A framework for evaluating when AI makes sense