No AI is now a feature
It's a weird timeline where not having built-in AI is now considered a feature.
Tag: ai
The barrier to generating content is almost zero with generative AI now
With the advent of AI, it's not possible to churn out content with almost no effort. Everyone can now generate AI content, or slop as it's known, at alarming rates, be it text, images, videos, etc. With agents, that generative AI content can be automated and posted on social media, a blog, a website, etc.
Call me crazy, but I don't want any of this AI-generated content. The more I see it, the less I want it.
OpenAI and Hugging Face hacking incident
The newly released independent METR-Redwood report is an eye-opener regarding the OpenAI and Hugging Face hacking incidents.
AI reductions
IA reductions are a real thing now. At an organization where my friend works, they are moving everyone off the $100 a month plan to the $20. The only teams that will remain with the $100/$200 are the Dev/OPs. This reduction will affect over 1,000 users.
AI crawlers and calendars
I was helping out a friend whose site got hit with a huge spike in traffic. This is a portfolio site. So no blog or anything like that. The entire site is six pages and static coded via Bootstrap. So why the 10,000x spike in traffic? It turns out the AI crawlers got stuck on his booking calendar in an endless loop. Once we blocked the AI crawlers, things returned to normal.
Troubleshooting with AI
As a SysAdmin, I am awed by AI's ability to troubleshoot error codes, logs, and bits of information. A simple copy and paste, and usually it will come back with a fix for the exact issue. Even pasting a screenshot will get you the desired results; there is no need to extract the text from the image. Gone are the days of searching Google or specific websites.
The best way to read the internet is via LLMs
No, seriously, you avoid all the ads, popups, and custom formatting, dark patterns, etc. Via an LLM, you get nice, clean, formatted content. Then again, the content you get might not be correct. So maybe only stick to things you really know? Not that misinformation doesn't exist via a search result. Right? Again, use your brain.
AI Copypasta
Please stop copying and pasting AI output. I'm begging you. I get it. You prompted it correctly and used one of those fancy skills. The output is relevant. The thing is, have you read it? Have you actually read all the output? Because if you didn't, then I don't want it. Is it still relevant? Is there anything you can trim?
Patching
AI seems to have lit a fire in, under, and all around securing software. The industry seems to be at odds with patching and maintaining software. As someone who had to patch a lot of infrastructure in the past, I don't see how teams are copping with this or how they are allocating time and resources to patch without inducing downtime. I see patching moving to a daily or hourly schedule in the near future with AI assistance. Especial since exploits aided by AI can be ready in hours. Soon that time will be minutes. I wonder how soon real-time patching will become the de facto standard?
- Microsoft Patches a Record 570 Security Flaws
- Linux kernel team publishes 432 CVEs in two days
- Oracle drops 1,449 security patches like it's the new normal
AI and the Three Laws of Robotics
With the recent OpenAI model breaking out of its sandbox and hacking Hugging Face and now others, I'm reminded that maybe we need something like Isaac Asimov's Three Laws of Robotics.
I'm also reminded that software comes with no guarantees, and most EULA and terms of service basically absolve everyone from everything, yet we all blindly accept them so we can play with the new shiny toys. I am aware that the U.S. government has thoughts about regulating AI models. Which I would imagine being similar to the regulations, reviews, and audits that are done by various federal branches that use software. But I think that's a slippery slope; if they start doing that to AI, then what comes next? All software?