I’m pretty sure that I’m not the only person wondering how the rapid emergence of AI will impact my work. I want to share my own thoughts and the way that I am responding, in my work, to the joint threat and wonder that is AI. Who knows, maybe it will trigger some debate?
Let’s start with a brief description of what I do for work. I create content, on the web, that I hope will be genuinely interesting to people. I do this because I’m fascinated by the content, and also because getting people to visit my website is the secret to many forms of monetization and I need to earn a living.
The content I’ve personally chosen to create is this - I watch TV cooking shows with an intent to figure out how they create those marvelous dishes, and then I set out to write about the ‘how’. I write about foods that we ordinary folks never think about putting together. I try to come up with recipes, and with cooking tips, that are useful for people with regular kitchens, regular pantries and regular budgets but who would still like to put irregular, wonderful food on the table.
So, that’s my bag. I typically generate about 50 pages of creative writing a week. I invent, on average, 5 new recipes a week. I adore my work.
How much of my work is better done by AI, and how quickly will that happen?
For this article, I think it will help if I divide the work I do into three buckets:
Writing about cooking shows, introducing lots of information
Cooking, and experimenting in the kitchen to re-create fascinating food for regular kitchens
Photographing and videoing the work that I do, so that I can pass it on to my readers
So, then, along comes AI.
Thinking about creating writing …
I am a firm fan of a podcast called “Hard Fork”. Funny, really, that I’m in the food business and I love a podcast with the word ‘fork’ in the title, but it’s nothing to do with eating. This is a technology podcast, and somehow it manages to be funny, and informative, and my Fridays are better for it. This week the hosts of Hard Fork were interviewing the CEO of Google’s Deep Mind and I learned that Gemini, Google’s AI platform, is now able to consume whole movies and, inside 2 minutes, answer any questions.
Well, yikes. I can tell you that it typically takes me about 10 hours to watch a one hour episode of a streamed cooking show and condense it down into interesting content. While I watch the show, I’m subconsciously brushing aside the more conventional issues and focusing on the bits of cooking that are both unusual and inspiring. I watched a program this morning where I ignored the cheese soup (been there, done that) and the fried pig’s tail (gotta be kidding me) but did single out the “potato crisps topped with a dome of bacon jelly”. Decades of cooking for friends tells me that the bacon jelly is something that many people will love so I want to try and create it. What I’m trying to tell you is that I report on what was cooked, but I also bring my own intuition to which parts of the transcript deserve a deeper dive.
Here’s what I think. AI can already transcribe and answer questions about a TV cooking show at a speed that makes me pointless. People browsing on the web to find out what that dish was that looked so delicious on The Great British Baking Show ™ will very soon not need to navigate their way to my little website to find out - Gemini, or ChatGPT, or Perplexity will tell them all about it in the blink of an eye. People will stop coming to my website for things like that before my two year old grandson is in school, I’m sure of it. No AI engine will need to pay me for a transcript when they can consume and spit out data at the speed of light.
BUT, I think that my aforementioned grandson will be an obnoxious teenager before the bots can replace my own instinct for knowing how to spot the allure of potato crisps with bacon jelly, over a deep-fried pig tail, and choose which one is most attractive to the typical diner.
What I have over AI is that there isn’t a large language model on the planet that has sat for many decades around kitchen tables, talking to beloved friends about what is good to eat, and what stirs our taste-buds. No AI device knows how much ‘umami’ matters, or that Brits generally love curry and Americans generally love sweet with savory. But I do. I know in my bones what people will choose off a menu and what they will avoid. I have won many, many cookery contests not just through cooking, but by knowing instinctively what foods combine well in the human yearning.
So, I’m going to slow down on transcribing and spend more time focusing on developing food fusions. I think that the bots will need me to do that for them for a long time yet.
Thinking about cooking and recipe development …
This one is pretty simple for me. I think my grandson will probably be a parent himself before robots can experiment with bacon, marrow bones, gelatin, agar-agar, pectin, etc and come up with the perfect jelly. Will there be chef-robots in my lifetime? Almost certainly, but I think they will all be pre-programmed to repeat cooking processes that have been carefully created by human beings.
Will AI be able to develop their own recipes by, independently, taking the pasta from recipe A and the sauce from recipe B and ‘inventing’ something new? Sure they will. But will they be able to intuit if their new recipe will have humans lined up down the street to try the new dish? Not for a long time, I think. Will they know if the set on a mousse just melts in the mouth? Will they know if the beet powder in the pasta fights with the clams in the sauce? Not for a very long time, I think.
So, I’m going to keep cooking and inventing. I think that the bots will need me to do that for them for a long time yet.
Thinking about food photography …
I think some of you will argue with me about this, but I’ll say it. I can’t compete. I have already given in and my website is littered with AI images.
When I develop a new recipe, it costs me a lot of money and a lot of time. I buy enough ingredients to cook the recipe at least six times over, because I know that I will keep re-testing until I trust in the dish. And by the fifth or sixth time of trying, I usually have the taste and the texture to my complete satisfaction, but I won’t necessarily have been thinking about the aesthetics. I might have made only enough sauce for me and my team to keep tasting, but not enough for a generous helping for photographic purposes. I might have the most delicious cake, but it was sliced unevenly in the process.
I can cook my new recipe AGAIN, this time with my focus on making it look perfect. That usually takes more than one attempt. Icing, piping, tweezing, cleaning … all takes time, and expertise. I confess that for me, perfection on a plate very much takes second fiddle to the taste and texture. And then there’s the paraphernalia of photography. Lighting, contrast, shadow. Clean kitchen, attractive backdrops. Ariel shots, portrait shots. Take a slice out of the cake and then try again - more shots. Editing, photo-shopping, publishing.
Or I can go to Bing or Dall-e and give AI a really good prompt. I asked Dall-e “please make me a photographic quality image of bite-size potato crisps, newly out of the fryer and glistening and puffed up, and topped with a dome of clear bacon jelly. Alongside, please show potatoes, rashers of bacon, and black pepper” and in less than 15 seconds I had this:
This might not be perfect, but it took seconds of my time and zero cost. When I think about what my readers want in the images I give them, I really believe that the answer is that they want pictures that help them figure out, at a glance, what I’m trying to say in words. Would my readers prefer that I spend a whole afternoon and a lot of money making a picture that is less than perfect, or give them an AI generated representation and put my time to things that a bot can’t do better?
There are many gifted photographers out there and I wholly admire what they’re doing. I’m not one of them, and for me, AI meets a need that I find it hard to ignore.
So, I’m going to keep cooking and inventing and leave AI to make my images. I simply have nothing to match their offering, so I’ve given up the fight. I’ll video or photograph my own work when it makes sense, but usually I’ll take the easy route.
And finally, the question I’m still trying to answer.
Is there any future for my website?
And I honestly think the answer is “not for the long term, no”. Let me try and explain why, and what I’m going to do about it.
My ideal followers are people who ask these sort of questions:
I want to cook with fish/macaroni/ham/something today. Can you find me a completely original recipe that I won’t have seen before but that makes my mouth water?
I saw someone on a tv cooking show cook some mousse with finger limes and yuzu. Can you help me make something similar?
I love watching The Great British Baking Show ™. This week they made croquembouche. How do I make that at home and what flavors can I use?
Can I really make risotto with potato instead of rice, and if so, how?
Today, I hope that those people will find my website, where I have to give them lots of context about why I’m developing unusual dishes, and what else I’ve seen on cooking shows. Personally, I find it really interesting, but (a) very soon, the chatbots are going to know more about the shows than I do and (b) who wants to try and discover my tiny presence on the internet when ChatGPT is in our face?
I think the writing is on the wall and while the AI platforms will need input from sites like mine, we’re basically going to feed the machine. Consumers of information will, inevitably and rightly, want to go to only one place to find everything they need. I hope, and trust, that the AI platforms will find a way to reward people like me who will continue to create unique content for the bots to serve up.
I already use Substack and ConvertKit to send my created material out to a list of people (please, sign up) and I don’t see that going away any time soon because I’m actively delivering it to my people. I just don’t think there’s much future in website-browsing for content any longer.
So, while I know that it’s an option for me to block the AI engines from crawling my site, I’m really doing the opposite. If I could plant a flag that says “pick me”, I would. If my website has a limited lifespan, then the sooner I get over the hump and feel that the work I’m doing has a long-term future, the better. I will continue to experiment with food, create new recipes, and occasionally photograph my work, but in the battle of AI versus creators, I feel like it’s time I stopped battling and tried to hold hands.
I’d love to get some feedback, please. Who else is doing what else? Who else thinks what else?
Jo
Comentários