Slokie Chef Instructions and Limitations


We'd appreciate it if you would disable ad blockers, since there are server costs to cover with this so we're keeping this free via ads. We may release a premium version down the road that allows users to create recipes from their own custom ingredient combinations.

To use the app, just enter some ingredients in the first box, or leave it blank if you prefer to focus more on nutritional requirements. Then, select your diet. The default is based on the Food and Drug Administration's posted minimum and maximum values based on a 2000 calorie diet. You may need to scale the resulting amounts based on your own body type. We'll try to add a feature to make this easier down the road. You can also select a high protein or a high fiber diet. Selecting either of these options will also sort the results based on fiber or protein, so even if you use the advanced options, you may want to ensure that either of these options are selected, based on what you'd like to see more of in your own diet. The final option opens the dietary constraints to their full ranges so that more results with more of the selected ingredients will be found. 


This project is still in its early alpha stage. You can peruse about 50 thousand pre-generated optimized ingredient mixes using the app below. Some already have ChatGPT recipes associated with them. Others don't. If you click the "Generate Meal Plan" button for a given list, and it appers to hang, just wait a minute or so... our system is just waiting for ChatGPT to finish it's piece. The next time you click that combination, it should respond immediately.

ChatGPT is obviously a huge step forward for machine learning. However, there are a lot of limitations. It quite frequently gives incorrect answers to questions, and one of the biggest struggles I've had to date with this project is ensuring that the ingredient lists that have optimized for nutritional value and the total quantity of food needed to fill you up get used. ChatGPT often only uses some of an ingredient for one particular meal, and then 'forgets' to use the rest of it. This means the calculated values for protein, fat, etc. will not be correct. Further, sometimes it will cook 'raw' ingredients, which may drastically change their nutritional contents, usually for the worse. Finally, this project currently only optimizes over the nutritional information associated with the sliders in the app. It does not factor in vitamins or minerals. Adding those in would increase the computational requirements for this project significantly, requiring the testing of tens of millions or even billions of ingredient combinations for each viable candidate.