John Kutay’s Post

Is AI progress slowing down? This (long) post is well worth reading in full – Arvind Narayanan, Benedikt Ströbl, and Sayash Kapoor do an excellent job of drilling into the current challenges with scaling AI models, why we shouldn’t trust industry insiders telling us that they have scaling AIs figured out, and why we are still years behind leveraging even the powers that AI bestows upon us today. > The furious debate about whether there is a capability slowdown is ironic, because the link between capability increases and the real-world usefulness of AI is extremely weak. The development of AI-based applications lags far behind the increase of AI capabilities, so even existing AI capabilities remain greatly underutilized. One reason is the capability-reliability gap --- even when a certain capability exists, it may not work reliably enough that you can take the human out of the loop and actually automate the task (imagine a food delivery app that only works 80% of the time). And the methods for improving reliability are often application-dependent and distinct from methods for improving capability. That said, reasoning models also seem to exhibit reliability improvements, which is exciting.

Serious question and I hope that I can get very concrete and serious answers: How did the emerging AI technologies increase the profits of the companies? Remember: concrete answers AND a comparison of how much is spent and how much was the ROI. It would also be good to compare rule-based algorithms to intelligent algorithms.

As I predicted before, products that want to be popularized to the majority of users will be optimized to reduce prices. AI is no exception and the birth of DeepSeek has caused a shock to the technology community, being evaluated as equal to and surpassing GPT 4 and Claude in many tasks with a 95% reduction in API costs and 75% reduction in memory, ... will accelerate the popularization of AI to the community. And now companies that do not know how to apply AI will gradually lose their competitive advantage and may be replaced in the future. Initially, AI can automate steps in the business process, but the destination of AI will be to transform the business processes.

DeepSeek AI is amazing on so many levels. Open source, only a few months old, has received only very little seed investments, better or equal to the best GPTs and comes at a fraction of the cost compared to what everyone is using right now and hence makes the commercial application much more appealing. Raises a lot of fundamental questions around the need for AI Capex spending (maybe not nearly as much needed?), market valuations (way too high?), etc.

AI The first thing to know about AI is to realize if this machine is smarter than you? Then make a decision from there. We are still smarter than technology. So let this be the way we think in the future. FOR THE WRITING IS ON THE WALL !!!!!!!!!!

The AI model race just got more interesting with Deepseek's breakthrough. Models are getting better AND cheaper. What's game changing isn't just in the models themselves. It's how companies that know their industries inside and out are using them. 1. Building specific workflows to solve their needs 2. Adding proprietary data 3. Solving real problems to deliver tangible value. The future belongs to those who build practical applications, not just better models. Customers get the best of both worlds - cheaper, stronger models AND solutions built just for them. The AI revolution isn't about who has the biggest model. It's about who uses it best 🔥

The AI industry is growing exponentially, and businesses that adopt it now will have a significant advantage in the years to come.

Just got around to reading Digital Trust and Safety Partnership's 'Best Practices for AI and Automation' guide. It's a really well done summary. Bookmark this resource and return to it. In Trust & Safety companies have been applying "AI", automated decision making, machine learning, computer vision, etc within critical business processes for years -- long before chat gpt got everyone talking about these technologies. So I may humbly suggest that there are lessons here for other fields as well. https://lnkd.in/eTy4S-i6

For companies: "A major driver of AI performance is not the technology itself, but who’s in the room making decisions about where, how, and when to employ AI"

AI is a massive crash in the making. That’s the potential takeaway from Daron Acemoglu, a professor at the Massachusetts Institute of Technology. As promising as AI may be, there’s little chance it will live up to Wall Street’s hype hype, he warns. By his calculation, only a small percentage of all jobs—a mere 5%—is ripe to be taken over, or at least heavily aided, by AI over the next decade. Good news for workers, true, but very bad for the companies sinking billions into the technology expecting it to drive a surge in productivity.