Another Crazy Day in AI: Why AI Won't Rule Everything
- Wowza Team
- Aug 5
- 4 min read
Hello, AI Enthusiasts.
Back in the swing of things? The AI world never left.
A new post from a longtime tech thinker says we’re in a “polytheistic AI” era. Instead of one dominant model, we’re watching multiple strong systems evolve at the same time... each with their own strengths, quirks, and limits. And yes, humans are still in the loop.
Meanwhile, a popular AI search tool is under fire for dodging site restrictions and crawling the web without permission.
And marketers now have a way to run AI tests inside live Google Ads campaigns—with real results and no messy duplication.
The week’s just warming up.
Here's another crazy day in AI:
Why AI isn’t taking over (yet)
Perplexity accused of evading crawl restrictions
Google Ads launches AI Max Experiments for faster testing
Some AI tools to try out
TODAY'S FEATURED ITEM: Beyond Single AI Dominance
Image Credit: Wowza (created with Ideogram)
Is the future of AI really about one supreme intelligence taking over, or are we heading toward something more like a diverse ecosystem of competing minds?
Technologist and writer Balaji Srinivasan makes the case that what we’re seeing now isn’t the rise of a single, all-powerful AI, but the emergence of many strong models developing in parallel. In a new blog published on his website, he outlines ten ideas that frame AI as something practical, decentralized, and still reliant on human input at key points. Drawing from his background as former CTO of Coinbase, general partner at a16z, and author of The Network State, Balaji offers a grounded look at what AI currently is, not what it might become someday. His central thesis revolves around what he calls “polytheistic AI”—the idea that we’re witnessing multiple strong AI models from different organizations creating a balance of power, rather than one system dominating everything else.
His Ten Observations
Several strong AI models are emerging at once from different companies with comparable capabilities, rather than one system achieving a massive lead
AI works as an intelligence booster where your skill at prompting and checking results determines what you actually get out of the system
Business costs are concentrating on the human parts of AI workflows—writing effective prompts and verifying outputs—while the computational processing gets cheaper
AI lets people become decent at new things like graphic design or programming, though specialists still provide the quality and polish that makes work truly professional
Visual tasks outperform text-based ones because people can immediately spot problems in images but need much more time to evaluate written content or code
Open-source models are spreading capabilities widely allowing smaller teams to access tools that were previously only available to large organizations
There's a middle ground for AI usage where some AI assistance helps productivity, but complete dependence often produces poor-quality results
AI faces genuine constraints in solving certain mathematical problems, managing costs, and operating without continuous human guidance
The polytheistic model Balaji presents stands in contrast to much of the current discourse around AI development. Instead of a winner-take-all scenario, he sees evidence of a more distributed future where multiple AI systems compete and complement each other. This competitive environment, combined with the practical limitations he identifies, suggests that human expertise will continue to play a crucial role—though perhaps in different ways than before. The emphasis on skills like effective prompting and careful verification indicates that people who learn to work well with AI tools may find themselves in increasingly valuable positions.
What makes Balaji's analysis particularly interesting is how it grounds AI development in current economic and practical realities rather than speculative futures. The constraints he describes—from API costs to verification overhead—point toward more strategic AI adoption rather than wholesale automation. His observations suggest we may be entering a period where different AI systems specialize in different areas, creating opportunities for users to choose the right tool for specific tasks rather than relying on a single all-purpose system. This distributed approach could lead to more resilient AI ecosystems and prevent the kind of technological concentration that many observers have worried about. The competitive landscape he describes may ultimately benefit users through continued innovation and diverse options, while the ongoing need for human oversight ensures that people remain central to how these powerful tools are deployed and managed.
Read the full blog here.
OTHER INTERESTING AI HIGHLIGHTS:
Perplexity Accused of Evading Crawl Restrictions
/Gabriel Corral, Vaibhav Singhal, Brian Mitchell, and Reid Tatoris, on The Cloudflare Blog
Cloudflare is calling out AI answer engine Perplexity for using undeclared crawlers and rotating IP addresses to bypass website no-crawl directives and access content without consent. While Perplexity declares a crawler identity, it appears to switch to stealth modes when blocked—violating crawling norms. Tests showed that even when robots.txt rules and firewalls were in place, Perplexity was still retrieving content from restricted domains. In response, Cloudflare has de-listed Perplexity as a verified bot and updated its defenses to block further stealth activity.
Read more here.
Google Ads Launches AI Max Experiments for Faster Testing
/Anu Adegbola, Paid Media Editor, on Search Engine Land
Google Ads has launched a new AI Max experiment tool that allows advertisers to test AI-powered features directly within live Search campaigns—without duplicating them. The feature splits campaign budgets 50/50 between control and experimental settings, enabling quicker, statistically valid insights. Advertisers can test features like Search Term Matching and Asset Optimization and choose whether to auto-apply results. The tool marks a shift in how AI is integrated into campaign workflows, aiming to streamline experimentation and reduce operational friction.
Read more here.
SOME AI TOOLS TO TRY OUT:
Communities – Like Discord, but for research paper discussions.
Lazy – One shortcut to capture and chat with your notes.
STORM – Turns any topic into a cited research report.
That’s a wrap on today’s Almost Daily craziness.
Catch us almost every day—almost! 😉
EXCITING NEWS:
The Another Crazy Day in AI newsletter is on LinkedIn!!!
Leveraging AI for Enhanced Content: As part of our commitment to exploring new technologies, we used AI to help curate and refine our newsletters. This enriches our content and keeps us at the forefront of digital innovation, ensuring you stay informed with the latest trends and developments.
Comments