AI
Sora Proves the AI Bubble Is Going to Burst
- Sora 2 allows copyrighted materials and celebrities’ likeness
- Estimated costs are $5 per video with no monetization plan, not to mention the resource utilization
- Why did Altman make a TikTok clone that no one really wanted? To keep the hype up.
- Tech is funding the equivalent of the Apollo space program every 10 months
- A large percentage of US GDP and the S&P 500 is tied to AI
- Financial analysts say that within 5 years, AI companies will have to bring in $2T in revenue
- Several studies (MIT, Becker Friedman Institute) report small or no gains with AI
- Although the models are getting bigger, the improvements are minimal in general. Many AI experts believe current approaches will not achieve AGI.
- AI CEOs and investors seem to agree this is a bubble
- (I personally disagree with Adam’s comment toward the end that no value has been brought by LLM tools. Sure, it didn’t deliver what was promised, but these tools still have value.)
The State of the AI Industry Is Freaking Me Out
This video talks about a diagram that depicts the major investors in AI and how they’re related.
Tech Billionaires Know the AI Bubble Will Burst
Here’s an additional analysis about the finances of AI, in particular circular deals and how OpenAI is inking multi-billion dollar deals with other companies despite not being profitable (and likely not being profitable until 2029). It also highlights that the 34% of the S&P 500 is made up of the Magnificent Seven (Nvidia, Meta, Google, Amazon, Apple, Microsoft, Tesla) and that US GDP growth has been primarily driven by AI despite many studies citing that these tools do not live up to the hype. The video also points out several economic downturns once a decade (Dot Com, toxic real estate assets, COVID-19) and how those affected particular industries harder; AI’s prevalence has the possibility of negatively impacting even more industries when the bubble bursts.
Why Agentic AI Projects Fail—and How to Set Yours Up for Success
“Gartner predicts that by the end of 2027, more than 40% of agentic AI projects will be canceled. The main reasons include escalating costs, unclear business value, and inadequate risk controls.” In short, adding agentic behavior to just anything isn’t the right call. Like most things with AI, AI is not the toolbox but one tool in your toolbox. Agentic AI excels in (1) complex, dynamic environments, (2) multi-step, multi-agent collaboration, and (3) high-value, low-frequency decisions.
Innovation
Why You Need Systems Thinking Now
This article explains breakthrough, design thinking, and systems thinking as modes of innovation. The first two tend to be short-term focused. “Systems thinkers reframe the definition of a complex problem iteratively to motivate stakeholders who may experience a system’s dysfunctions differently.”
Software
Software Design Is NOT Software Architecture (& That’s A Good Thing!)
- Architecture = design decisions that are hard to change
- Design = decisions about the low-level details; inexpensive to change
- Modularity helps to minimize the blast radius
- Automated tests help you understand how change affects the system
- Refactoring skills help you make changes safely