Episode published by​ ​The Retort​

Carl Brown (Internet of Bugs): The Media’s and Influencers’ Narrative about AI Vibe Coding is Nonsense, Dangerous, and Misleading

Carl Brown of Internet of Bugs returns to discuss the oversold hype around AI vibe coding versus reality, and what the recent Adobe story really shows us.

You may have seen the recent headlines about the developer Brandon Thomas and co-founder Michael Caffrey using Anthropic’s Claude Opus 5.5 to build clean-room, open-source clones of seven major Adobe programs written in Rust.

As we have been calling out on this podcast, the media and certain podcasters and influencers suggested that, thanks to AI vibe coding, this small team could “decompile” and recreate Adobe’s major software products and threaten Adobe’s future forever. Obviously, this is not true!

Carl examined this claim and explained why early prototypes are not the same as polished, production-ready replacements as people might think. His experience testing an AI-built editing tool showed missing features and familiar workflows that simply were not available.

The discussion explores why people and businesses continue to rely on established software like Adobe, even when alternatives exist do exist and if AI-powered vibe coding has any place in the real world.

More ecritially, we discussed the why reliability, existing workflows, support, and the career value of learning widely used tools still all matter. It is not just about price or feature lists, certainly not in the enterprise world.

We also consider how subscription fatigue and customer frustration can create opportunities for competitors and new startups.

Carl distinguishes reverse engineering – recreating software behaviour from observation and specifications – from decompiling, which attempts to turn compiled software back into readable code. He argues that media coverage often blurs the terms, making a story sound more dramatic than the technical reality supports.

The broader takeaway from our discussion is that AI can amplify skilled developers and speed up early work. However, complex debugging, reliable products, and sound judgment still require expertise that AI can never replicate. We apply this same call for nuance to AI security, SaaS, hiring, and media coverage. You must look past the headline and ask what was actually built and how well it works.

Join the conversation and tell us if you think vibe coding creates more problems than solutions, especially if used by individuals who have never coded or built software and technology businesses.

• ​00:00:00 – Preview
• ​00:02:40 – Independent media faces serious challenges, and your support matters more than ever in the age of AI slop, clickbait and misinformation
• ​00:03:30 – How the media overly exaggerated the AI-built Adobe alternatives and why they are not real replacements
• ​00:09:14 – Why, despite media and influencer hype, people will stick with Adobe’s core competency — and the business case for software
• ​00:15:22 – How Adobe’s shift from boxed software to subscriptions became the epitome of rip-off pricing, and when subscriptions make sense
• ​00:22:17 – Why AI amplifies developer capability rather than replacing it, and how it helps with easy tasks while complicating hard ones
• ​00:31:13 – What decompiling means, whether the team decompiled or reverse-engineered Adobe, and why it matters
• ​00:41:59 – How AI has made software piracy and intellectual property theft easier, examining Sony’s recent PS5 jailbreaks
• ​00:47:29 – How subscription pricing killed real ownership and fuelled the rise of piracy, with the mantra ‘if buying isn’t owning, pirating isn’t stealing’
• ​00:50:35 – Are we witnessing the death of downloadable software and apps, and how this contradicts the media narrative of AI vibe coding killing the SaaS market?
• ​00:57:12 – Winners and losers in the software market, and Canva’s acquisition of Affinity as an example of a new hybrid software service model
• ​01:05:35 – How AI hype and media incentives exploit customer frustration with predatory SaaS and streaming subscription pricing and zero ownership
• ​01:12:02 – Can AI vibe coding help organisations ditch paid-for software for open source by hiring internal developers and saving money?
• ​01:16:17 – Calling out media outlets, podcasters and influencers spreading nonsense like ‘AI agents building their own governments’
• ​01:30:38 – How the current AI hype cycle suits bullshittery, laziness and poor-quality human output, looking at the disastrous AI-powered hiring market
• ​01:37:04 – The emerging fallacy of deploying AI models to solve the endless problems caused by AI models, looking at cybersecurity as an example