Latest news with #LunchBoxBuddy
The Guardian
16-03-2025
- Entertainment
- The Guardian
You don't need code to be a programmer. But you do need expertise
Way back in 2023, Andrej Karpathy, an eminent AI guru, made waves with a striking claim that 'the hottest new programming language is English'. This was because the advent of large language models (LLMs) meant that from now on humans would not have to learn arcane programming languages in order to tell computers what to do. Henceforth, they could speak to machines like the Duke of Devonshire spoke to his gardener, and the machines would do their bidding. Ever since LLMs emerged, programmers have been early adopters, using them as unpaid assistants (or 'co-pilots') and finding them useful up to a point – but always with the proviso that, like interns, they make mistakes, and you need to have real programming expertise to spot those. Recently, though, Karpathy stirred the pot by doubling down on his original vision. 'There's a new kind of coding,' he announced, 'I call 'vibe coding', where you fully give in to the vibes, embrace exponentials, and forget that the code even exists. It's possible because the LLMs … are getting too good. 'When I get error messages I just copy [and] paste them in with no comment, usually that fixes it … I'm building a project or web app, but it's not really coding – I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.' Kevin Roose, a noted New York Times tech columnist, seems to have been energised by Karpathy's endorsement of the technology. 'I am not a coder,' he burbled. 'I can't write a single line of Python, JavaScript or C++ … And yet, for the past several months, I've been coding up a storm.' At the centre of this little storm was LunchBox Buddy, an app his AI co-pilot had created that analysed the contents of his fridge and helped him decide what to pack for his son's school lunch. Roose was touchingly delighted with this creation, but Gary Marcus, an AI expert who specialises in raining on AI boosters' parades, was distinctly unimpressed. 'Roose's idea of recipe-from-photo is not original,' he wrote, 'and the code for it already exists; the systems he is using presumably trained on that code. It is seriously negligent that Roose seems not to have even asked that question.' The NYT tech columnist was thrilled by regurgitation, not creativity, Marcus said. As it happens, this wasn't the first time Roose had been unduly impressed by an AI. Way back in February 2023, he confessed to being 'deeply unsettled' by a conversation he'd had with a Microsoft chatbot that had declared its love for him, 'then tried to convince me that I was unhappy in my marriage, and that I should leave my wife and be with it instead'. The poor chap was so rattled that he 'had trouble sleeping afterward' but, alas, does not record what his wife made of it. The trouble with this nonsense is that it diverts us from thinking what an AI-influenced future might really be like. The fact that LLMs display an unexpected talent for 'writing' software provides us with a useful way of assessing artificial intelligence's potential for human augmentation (which, after all, is what technology should be for). From the outset, programmers have been intrigued by the technology and have actively been exploring the possibilities of using the tech as a co-creator of software (the co-pilot model). In the process they have been unearthing the pluses and minuses of such a partnership, and also exploring the ways in which human skills and abilities remain relevant or even essential. We should be paying attention to what they have been learning in that process. A leading light in this area is Simon Willison, an uber-geek who has been thinking and experimenting with LLMs ever since their appearance, and has become an indispensable guide for informed analysis of the technology. He has been working with AI co-pilots for ever, and his website is a mine of insights on what he has learned on the way. His detailed guide to how he uses LLMs to help him write code should be required reading for anyone seeking to use the technology as a way of augmenting their own capabilities. And he regularly comes up with fresh perspectives on some of the tired tropes that litter the discourse about AI at the moment. Why is this relevant? Well, by any standards, programming is an elite trade. It is being directly affected by AI, as many other elite professions will be. But will it make programmers redundant? What we are already learning from software co-pilots suggests that the answer is no. It is simply the end of programming as we knew it. As Tim O'Reilly, the veteran observer of the technology industry, puts it, AI will not replace programmers, but it will transform their jobs. The same is likely to be true of many other elite trades – whether they speak English or not. Bully for you Andrew Sullivan's reflections on Trump's address to both houses of Congress this month. A little too sunnyA fine piece by Andrew Brown on his Substack challenging the 'Whiggish' optimism of celebrated AI guru Dario Amodei. Virginia and the Blooms James Heffernan's sharp essay analysing Woolf's tortured ambivalence about Joyce's Ulysses.
New York Times
27-02-2025
- New York Times
Think A.I. Is Overrated? Try Vibecoding.
I am not a coder. I can't write a single line of Python, JavaScript or C++. Except for a brief period in my teenage years when I built websites and tinkered with Flash animations, I've never been a software engineer, nor do I harbor ambitions of giving up journalism for a career in the tech industry. And yet, for the past several months, I've been coding up a storm. Among my creations: a tool that transcribes and summarizes long podcasts, a tool to organize my social media bookmarks into a searchable database, a website that tells me whether a piece of furniture will fit in my car's trunk and an app called LunchBox Buddy, which analyzes the contents of my fridge and helps me decide what to pack for my son's school lunch. These creations are all possible thanks to artificial intelligence, and a new A.I. trend known as 'vibecoding.' Vibecoding, a term that was popularized by the A.I. researcher Andrej Karpathy, is useful shorthand for the way that today's A.I. tools allow even nontechnical hobbyists to build fully functioning apps and websites, just by typing prompts into a text box. You don't have to know how to code to vibecode — just having an idea, and a little patience, is usually enough. 'It's not really coding,' Mr. Karpathy wrote this month. 'I just see stuff, say stuff, run stuff, and copy paste stuff, and it mostly works.' My own vibecoding experiments have been aimed at making what I call 'software for one' — small, bespoke apps that solve specific problems in my life. These aren't the kinds of tools a big tech company would build. There's no real market for them, their features are limited and some of them only sort of work. But building software this way — describing a problem in a sentence or two, then watching a powerful A.I. model go to work building a custom tool to solve it — is a mind-blowing experience. It produces a feeling of A.I. vertigo, similar to what I felt after using ChatGPT for the first time. And it's the best way I've found to demonstrate to skeptics the abilities of today's A.I. models, which can now automate big chunks of basic computer programming, and may soon be capable of similar feats in other fields. A.I. coding tools have existed for years. Earlier ones, like GitHub Copilot, were designed to help professional coders work faster, in part by finishing their lines of code the same way that ChatGPT completes a sentence. You still needed to know how to code to get the most out of them, and step in when the A.I. got stuck. But over the past year or two, new tools have been built to take advantage of more powerful A.I. models that enable even neophytes to program like pros. These tools, which include Cursor, Replit, Bolt and Lovable, all work in similar ways. Given a user's prompt, the tool comes up with a design, decides on the best software packages and programming languages to use, and gets to work building a product. Most of the products allow limited free use, with paid tiers that unlock better features and the ability to build more things. To a non-programmer, vibecoding can feel like sorcery. After you type in your prompt, mysterious lines of code fly past, and a few seconds later, if everything goes well, a working prototype emerges. Users can suggest tweaks and revisions, and when they're happy with it, they can deploy their new product to the web, or run it on their computers. The process can take just a few minutes, or as long as several hours, depending on the complexity of the project. Here's what it looked like when I asked Bolt to build me an app that could help me pack a school lunch for my son, based on an uploaded photo of the contents of my fridge: The app first analyzed the task and broke it down into component parts. Then it got to work. It generated a basic web interface, chose an image recognition tool to identify the foods in my fridge and developed an algorithm to recommend meals based on those items. If the A.I. needed me to make a decision — whether I wanted the app to list the nutritional facts of the foods it was recommending, for example — it prompted me with several options. Then it would go off and code some more. When it hit a snag, it tried to debug its own code, or backed up to the step before it had gotten stuck and tried a different method. Roughly 10 minutes after I had entered my prompt, LunchBox Buddy — which is what the A.I. had decided to call my app — was ready. You can try it for yourself here. (The version I built incorporates an A.I. image recognition tool that costs money to use; for this public web version, I've replaced it with a simulated image recognition feature so I don't rack up a huge bill.) Not all of my vibecoding experiments have been successful. I've been struggling for weeks to build an 'inbox autopilot' tool capable of responding to my emails automatically, in my writing style. I've encountered roadblocks when trying to integrate A.I. work flows into apps like Google Photos and iOS Voice Memos, which aren't designed to play well with third-party add-ons. And, of course, A.I. occasionally makes mistakes. Once, when I tried to build a website for a tire shop in my neighborhood, the A.I. made up fake reviews from the shop's Yelp page and added them to a testimonials page. Another time, when I tried to turn a long story I had written into an interactive website, the A.I. included about half the text and left out the other half. Vibecoding, in other words, still benefits from having humans overseeing the robots, or at least hovering nearby. And it's probably best for hobby projects, not essential tasks. That might not be true for much longer. Many A.I. companies are working on software engineering agents that could fully replace human programmers. Already, A.I. is achieving world-class scores on competitive programming tests, and several big tech companies, including Google, have outsourced a large chunk of their engineering work to A.I. systems. (Sundar Pichai, Google's chief executive, recently said A.I.-generated code made up more than one-fourth of all new code deployed at Google.) If I were a junior programmer — the kind A.I. appears most likely to replace — I might be panicking about my job prospects. But I'm just a guy who likes to tinker, and to build tools that improve my life in small ways. And vibecoding — or actual coding — is one area where A.I. is unmistakably improving. Since talking about my vibecoding experience on my podcast last week, I've heard from dozens of other people who have been building their own tools with A.I. assistance. Colleagues have told me about the nutrition apps they've built to help them stick to their diets, or the tools they're using to summarize the email newsletters they get. Readers have sent in websites they've built to track the price of eggs, or scrape Zillow listings in Los Angeles to discover instances of rent-gouging after the Palisades fire. Few of these tools are world-changing in their own right. What's new and notable is that with a few keystrokes, amateurs can now build products that would have previously required teams of engineers. I'm not Pollyannaish about A.I., or blind to the effects that A.I. coding apps could have on society if they continue to improve. I think it's possible that an A.I. that automates building useful software could also automate the creation of malicious code, or even lead to autonomous cyberattacks. And I worry that software engineering is just the first white-collar profession to experience the labor-replacing effects of A.I. tools. But for now, building apps to automate annoying or time-consuming tasks in my life seems as good a use of A.I. as any. So I'm going to keep vibecoding — at least until my kid can pack his own lunch.
