![\"Vibe](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/10\/869efbde1918b700c16ac988414d1738a35f07a2.jpeg\")
<\/p>\nAll of this sounds promising, even magical. But Dorsey\u2019s app spree sums up a lot of what\u2019s wrong with Silicon Valley. Vibe coding seems to be largely about solving problems that don\u2019t exist, then charging users $4.99 per month for the privilege of having access to a tech scion\u2019s brain fart.<\/p>\n
Now, this trend is metastasising beyond app development. Microsoft is betting big on what it calls \u201cvibe working\u201d \u2013 bringing the same prompt-based approach to Office apps.<\/p>\n
The company\u2019s new Agent Mode in Excel and Word, along with its Office Agent powered by Anthropic models, promises to generate complex spreadsheets, documents, and PowerPoint presentations from simple chatbot prompts. Microsoft is literally marketing the ability to produce \u201cwork, quite frankly, that a first-year consultant would do, delivered in minutes\u201d.<\/p>\n
If that sounds like a red flag to you, you\u2019re not alone. The same impulse that drives vibe coding \u2013 racing to get something out there with minimal human oversight \u2013 is now being applied to the documents, spreadsheets, and presentations that power businesses worldwide. And Microsoft admits its Agent Mode in Excel has an accuracy rate of just 57.2 per cent, well below human accuracy of 71.3 per cent. That\u2019s not a reassuring margin of error when dealing with business-critical data.<\/p>\n
It\u2019s about racing to get something out there \u2013 no matter how rushed or how little demand \u2013 then taking time to correct the mistakes later (if bothering to correct them at all). Vibe coding lacks the security requirements or rigorous testing that normally are crucial aspects of the development process.<\/p>\n
<\/p>\n
Vibe coding is a new trend in which engineers use AI and natural language to autonomously make apps.Credit: Bloomberg<\/p>\n
Take Dorsey\u2019s other app, Bitchat. One of its main points of difference is that it\u2019s supposed to be private and secure. Except within days of being available, coders figured out how to impersonate other users \u2013 a pretty major security flaw.<\/p>\n
More than once, we\u2019ve learnt the hard way that we cannot trust Silicon Valley to do the right thing when left to its own devices. And that\u2019s even more true for apps that are created by AI.<\/p>\n
For example, how can an app developer be trusted to plug a security hole in an app that a human being didn\u2019t even code? And when Microsoft describes its Excel features as \u201cwork that a first-year consultant would do,\u201d should we really be celebrating the automation of critical business documents by AI that performs worse than humans?<\/p>\n
Loading<\/p>\n
Both Australian and global regulators have already got their hands full trying to unpick some of the tech industry\u2019s worst mistakes. The Albanese government\u2019s teen social media ban is an attempt to reverse damaging psychological and social impacts from platforms that have long been proven to damage young people\u2019s body image and self-worth.<\/p>\n
The problem here isn\u2019t Jack Dorsey himself, or even Microsoft. It\u2019s fair to say the softly spoken mogul is more thoughtful and considered than many of his Palo Alto peers, who tend to be more interested in sheer power and racking up daily active users than safety. And Microsoft, to its credit, has taken a more gradual approach to AI in Excel given the critical business data it handles.<\/p>\n
But \u201cvibe coding\u201d an app and releasing it haphazardly for the world to use, without the usual checks, tests or deliberations, is a recipe for disaster. And expanding this philosophy to \u201cvibe working\u201d \u2013 where AI generates the actual substance of our professional output \u2013 raises the stakes even higher.<\/p>\n
![\"Jack](\"https:\/\/www.newsbeep.com\/nz\/wp-content\/uploads\/2025\/10\/cfc225fcc09a1a5d1ee89f7f530da2e82daa8515.jpeg\")
<\/p>\n
Jack Dorsey may not be as cavalier as some others, but his plans for \u201cvibe working\u201d should give us pause.Credit: Getty<\/p>\n
It speaks to the general unseriousness with which the Valley treats its power time and again.<\/p>\n
What we actually need from these ultra-wealthy and mega-powerful few is a vibe shift \u2013 away from recklessness and negligence, and towards care, thoughtfulness and apps and technologies that serve us rather than merely distract us or worse, prey on us.<\/p>\n
At present, we can rely on AI to take care of menial, mundane tasks that will free us up to do more interesting things. But vibe coding our apps \u2013 or vibe working our spreadsheets and presentations \u2013 on a Sun Day or any other day, isn\u2019t the answer.<\/p>\n