Why People Trust (or Don’t Trust) AI Recipes in 2025: A Kitchen Psychology Guide
The numbers tell an interesting story about AI recipes. Only 18% of people use AI-helped cooking while 82% stay away from it . AI recipes are spreading fast across digital platforms, but trust remains the biggest roadblock to wider adoption. This became clear when Tasty did a blind taste test between a GPT-3 chocolate cake recipe and one from a professional food writer. The human recipe won with better flavor and texture .
People’s trust in AI recipes changes based on what they’re cooking. Research shows that for basic dishes, people trust AI recipes just as much as traditional ones. But their trust drops when AI tries to create new and innovative recipes . The relationship with AI cooking is not simple. People tend to trust it more when they believe it’s accurate and feel good about technology in general. Fear of AI makes them less likely to use these recipes . This lack of trust is more than just a preference issue. AI recipes can sometimes be dangerous. One AI system created a harmful chemical mix when asked for a water-based recipe . This detailed look shows what makes people trust AI recipes, why they choose AI for cooking, and how developers can build AI cooking apps that people will trust.
Why trust matters in AI-generated recipes
Trust is the life-blood of AI’s future in culinary applications. It will determine if people embrace or reject AI-generated recipes. Research shows that trust shapes how consumers make decisions about new food technologies [1]. This relationship goes beyond simple preferences to include health concerns, cultural identity, and food safety.
The rise of AI in the kitchen
AI has grown faster across the digital world. You’ll find AI-generated recipes on Instagram, TikTok, and Pinterest [2]. These systems learn from huge collections of recipes and food images to create new dishes. AI has become a helpful kitchen assistant over the last several years through several key uses:
- It suggests recipes based on what’s in your kitchen, which helps with dietary restrictions and limited ingredients [3]
- It helps manage inventory and reduces food waste through up-to-the-minute data analysis [3]
- It automates basic tasks like chopping, slicing, and mixing [3]
It also helps professional chefs be more creative. Many culinary experts now use AI tools to break through creative blocks, try new ingredient combinations, and adapt recipes for special diets or cultural twists [3]. To name just one example, see how a culinary director at Velvet Taco created a weekly special with ChatGPT’s help. This shows how AI can spark creativity without replacing human expertise [1].
Why people hesitate to trust AI for food
Even with these advances, many people remain skeptical. A survey by Ingredient Communications found that 83% of respondents want food companies to tell them if AI helped design or make a product [4]. About 44% worry that AI-created food might not be safe [4].
People hesitate because:
They don’t trust AI’s accuracy. AI-generated recipes often miss quantities, mix up ingredient orders, or combine ingredients that don’t work together [3]. People feel anxious about AI’s unpredictability, which makes them focus on risks instead of benefits [1]. Unlike human food creators who test their recipes, AI systems don’t take responsibility for their output [2].
Research shows different types of trust work together. Rational trust based on accuracy creates the foundation for emotional trust and preference [1]. Social norms and quality standards build rational trust, while perceived risks hurt emotional trust and buying decisions [1].
The stakes: health, taste, and safety
Bad AI recipes can do more than ruin dinner – they can create serious health and safety issues. AI can’t taste, feel, or smell food, which makes it hard to develop good recipes [5]. This creates several big risks:
- Safety hazards: AI sometimes suggests dangerous ingredient combinations. A supermarket’s AI chatbot in New Zealand gave recipes that would have created toxic chlorine gas [3]. AI-generated mushroom foraging books included deadly mistakes [1].
- Food quality issues: Human recipes beat AI in blind taste tests. Tasty compared a GPT-3 chocolate cake recipe with one from a food writer. Tasters liked the human recipe better because it had better flavor and texture [5].
- Technical inadequacies: AI recipes often skip important details that experienced cooks would include. They miss proper ingredient order, cooking times, or visual clues that help beginners succeed [1]. This especially hurts new cooks who need the most help but can’t spot recipe problems.
AI recipe systems pull information from websites without checking if they’re reliable [6]. They don’t deal very well with important details like pressure canner types, elevation needs, or proper headspace for canned goods. These mistakes can make preserved foods unsafe [6].
The importance of trusting AI recipes goes beyond convenience – it affects our health, enjoyment of food, and safety standards.
The psychology of trust in AI recipes
“Trust is not just a technical problem—it is deeply human. And trust in AI will not (and should not) be automatic. Just like with people, it has to be earned, nurtured, and sometimes repaired.” — Aparna Chennapragada, Former Chief Product Officer, Robinhood; ex-Google VP, Product
The way people trust AI recipe generators shows a fascinating mix of logical thinking and emotional responses. We need to understand this relationship to explain why people either accept or reject ai generated recipes in different situations. Research shows that trust isn’t just one thing – it’s made up of different psychological parts that grow in specific ways.
Cognitive trust vs affective trust
People trust ai recipe systems in two different ways: through thinking and feeling. Cognitive trust comes from what we know and helps us predict if AI can cook well [1]. This means people trust AI-made food’s safety, nutrition, and taste based on what they know about AI technology [1]. Affective trust, on the other hand, is all about emotions – it’s when people feel safe and confident about using the technology [1].
These two types of trust work together in a specific way. Studies show that thinking-based trust leads to feeling-based trust [1]. Then, people who really believe in ai in the kitchen are more likely to feel good about using these systems [1]. This explains why users might first understand why an AI recipe makes sense before they feel comfortable following its suggestions.
How trust is built over time
People don’t instantly trust AI cooking tools – it builds up over time. The thinking part comes first [1]. As people have more good experiences with ai recipe app systems, their logical trust turns into emotional trust [1]. This is just like how we learn to trust other people – our experiences shape how much we trust them [4].
People trust recipes more when they match what they already know about cooking, rather than how hard the recipe is to make [4]. That’s why AI-made traditional recipes get as much trust as cookbook ones, while new AI recipes get much less [4]. Research found people trusted AI and cookbook instructions equally for classics like Negroni and lasagna [4]. But when AI tried to make new versions of these recipes, people didn’t trust them as much [4].
The role of past experiences
Your previous experiences with ai for recipes really affect how much you’ll trust them later. Research shows good experiences make people more likely to use AI cooking systems, while bad ones make them avoid them [3]. This works the same way we trust human chefs – when they do well, we expect them to keep doing well.
Past experiences matter beyond just personal ones – they include cultural and social factors too. Researchers point out that people trust chefs because they help keep cooking traditions alive [4]. This makes it hard for AI recipe systems to compete since they don’t have the same history as human experts.
All the same, building trust looks toward the future. Teaching programs and practice runs that give people good first experiences with AI can help adjust how they see risks and build their confidence [3]. This shows that while past experiences matter, trust and how we see things play bigger roles in deciding whether to use ai generated recipes [3].
Key factors that influence trust in AI for recipes
Research shows that trust in ai recipe generators depends on three connected factors that affect how users see and use these technologies. These factors help explain the big gap between people who quickly adopt and those who doubt AI cooking tools in the digital world.
Perceived accuracy of AI-generated recipes
How accurate people think ai recipes are forms the foundation of trust. Users who believe AI systems create reliable recipes tend to trust these technologies more [1]. This belief plays a key role in whether people trust AI cooking suggestions as much as traditional recipes.
Many accuracy problems still exist with ai generated recipes:
- Wrong ingredient amounts or missing measurements
- Steps listed in the wrong order
- Unrealistic cooking times leading to raw food
- Ingredients that don’t work together
These problems affect what people think, especially after some dangerous AI suggestions. A supermarket’s AI chatbot in New Zealand once suggested recipe ingredients that would have created toxic chlorine gas [7]. AI has also made dangerous mistakes in mushroom guides by giving deadly advice about which ones are safe to eat [7].
Safety issues aside, accuracy problems lead to disappointing results. Tasty tested a GPT-3 recipe against a professional food writer’s chocolate cake. The AI version was edible but not as good – it turned out too sweet and dry [8]. Such results make people doubt if AI can reliably create good recipes.
General attitudes toward AI
What people think about AI technology as a whole strongly affects their willingness to trust ai in the kitchen. Studies show different age groups and education levels have different views. Younger, educated people usually feel more comfortable with AI cooking help [2]. People who know more about technology tend to trust AI more.
Trust levels change based on what you’re cooking. People accept AI more when it helps with practical tasks like using leftovers or checking what’s in the pantry [2]. They trust it less when cooking for important events like dinner parties where mistakes matter more [2].
A recent survey found 83% of respondents want food companies to tell them if they used AI in making their products [9]. People clearly want to know when AI plays a role in their food.
AI anxiety and fear of errors
Fear of AI—worry about unexpected results and possible risks—makes people trust ai for recipes less [1]. This worry usually focuses on specific problems rather than a general fear of technology.
Studies show that people who worry more about AI see fewer benefits and more risks, which leads to less trust [1]. Past mistakes with AI food advice have made this worse. Wrong portion sizes, incorrect calorie counts, and bad ingredient combinations have made people more careful [1].
Trust makes a big difference in how people use AI cooking tools. People with higher trust levels (+1 standard deviation) are 5.44 times more likely to use AI cooking apps than those who don’t trust them [2]. This shows how accuracy, attitudes, and anxiety work together to determine if people will welcome ai recipe app technology into their kitchens.
When people trust AI in the kitchen—and when they don’t
The way users trust ai recipe systems depends on context. They trust AI differently based on the cooking situation. Users naturally assess risks before they follow AI’s cooking guidance. Their decisions show how they balance benefits and risks carefully.
Low-stakes scenarios: using leftovers
People trust ai generated recipes more when they face everyday cooking challenges that don’t have major social or health risks. Cognitive load theory suggests these situations work best with AI help because they reduce mental effort without much risk. Here’s where ai recipes get the most trust:
People trust AI most when they need to use up leftover ingredients. They like how AI can suggest creative ways to combine available ingredients. This matches AI’s strength in spotting patterns and connecting information from databases. Daily meal prep also gets high trust scores, especially from busy users who want quick solutions rather than perfect meals.
Recipe modification is another area where users trust AI. People feel confident using ai recipe app suggestions to adjust recipes for allergies, restrictions, or nutrition goals. These tasks need systematic substitution rules that fit AI’s analytical strengths. But users still check AI’s suggestions against what they know, which shows they see AI as a partner rather than taking over completely.
High-stakes scenarios: dinner parties and guests
Trust drops by a lot during high-stakes cooking where social pressure, safety worries, or emotional value make AI mistakes more costly. People avoid using ai in the kitchen most during dinner parties and special occasions. This comes from complex psychological factors, including worry about impressions and emotional investment in these events.
Users also trust AI less when cooking unfamiliar cuisines or technically challenging dishes. They hesitate to rely on AI for dishes that need cultural understanding or precise techniques – things they connect more with human expertise. This doubt grows stronger when cooking unusual ingredient combinations that might be unsafe, since AI can’t taste, smell, or see potential problems.
Professional and serious home cooks don’t trust ai for recipes much when creativity and personal expression matter. This resistance shows how cooking shapes identity, where recipes become personal statements rather than just instructions.
Functional vs emotional cooking contexts
A reliable way to predict ai recipe trust patterns comes from looking at functional versus emotional cooking. Functional cooking – which focuses on efficiency, nutrition, and problem-solving – matches what users think AI does well with data and optimization. These situations highlight cooking’s practical side, like saving time or hitting nutritional targets.
Emotional cooking focuses on human elements that users value more:
- Cultural authenticity and traditions
- Emotional comfort and nostalgia
- Creative expression and innovation
- Social bonds through shared cooking experiences
This difference explains why people trust AI for meal planning, inventory tracking, and nutrition analysis (functional tasks) but prefer humans for holiday meals, cultural dishes, and creative cooking (emotional tasks). This pattern shows up clearly in mixed situations like weeknight family dinners, where users let ai recipe systems help with practical tasks while they keep control of the emotional aspects.
People’s trust in AI cooking systems comes from a complex evaluation that considers both risks and emotional importance, not just their general views on technology.
The impact of recipe type: traditional vs innovative
Recipe complexity plays a key role in how people accept ai recipes. The differences between traditional and creative dishes stand out clearly. Research shows that the recipe type—whether conventional or new—affects trust levels whatever the technology behind the ai recipe app.
Why traditional recipes are more trusted
Research about trust in culinary AI shows an interesting pattern: people trust AI-generated and human-created versions of traditional recipes about the same. Tests with classic dishes like the Negroni cocktail or lasagna Bolognese showed no significant differences in trust between AI and human-made instructions [10]. This shows that ai generated recipes can earn the same trust as human chefs for familiar dishes.
The reason lies in recipe matching—how well the recipe’s steps and ingredients line up with what cooks already know [4]. Traditional recipes use ingredients people recognize and follow expected cooking methods. This triggers familiar cooking patterns that help build trust. Research shows that the recipe’s match with known cooking rules matters more than how complex it seems [4].
How innovation triggers skepticism
On the flip side, creative AI recipes face much more doubt. Tests with unique variations like white Negroni or tomato-jam bacon lasagna showed much lower trust in AI versions compared to similar recipes from human chefs [10]. People had less faith in the taste, smell, and were less willing to try these recipes [6].
The doubt about creative ai in the kitchen comes from several connected factors:
- Perceived creativity limitations: People question if AI can create without tasting
- Safety concerns: New ingredient combinations make people worry more
- Cultural context deficit: New recipes need cultural knowledge that AI lacks
- Technical credibility gap: Creative recipes need more than just computing power
University of Oxford research shows this trust gap exists even when ai for recipes look good—people liked photos of AI-made food until they learned AI made them [5]. This shows how people value creativity more when it comes from humans rather than AI.
The role of familiarity in food choices
Familiarity serves as the life-blood of food acceptance psychology. This explains why people trust traditional and innovative ai recipes differently. Food neophobia—avoiding unfamiliar foods—affects food choices across all age groups [1]. Kids, who usually show stronger food neophobia, like AI-made recipes less than those from human chefs because they think AI recipes are too unusual [1].
Familiarity’s effect goes deeper than just preferences. Chefs hold trusted positions because they keep and share cooking traditions [4]. This makes it harder for AI systems to introduce new dishes that lack history or cultural background.
All the same, these trust patterns keep changing. Just like ChatGPT faced resistance before people accepted it, users slowly trust ai recipes more as they succeed with AI cooking helpers [4]. This rise suggests that while recipe type sets trust limits now, people might accept innovative ai generated recipes more as they have good experiences with AI cooking systems.
Who trusts AI recipes? Demographic and psychographic insights
Demographics and psychology play a major role in determining who uses ai recipes in their cooking. Studies show clear patterns among different groups of people. Their comfort with technology and personality traits determine how much they trust these systems.
Younger vs older users
Age makes a big difference in who uses ai recipe tools. The numbers tell an interesting story – 74% of people aged 18-24 have tried AI cooking tools. This number drops steadily as age increases [11]. ChatGPT serves as a perfect example, where 69% of users between 18-30 use it. The numbers drop substantially for older age groups [12].
The findings about younger users might surprise you. Research shows that 18-24-year-olds worry more about the missing human element in ai in the kitchen than any other group. Their concern runs almost five percentage points higher than other age groups [11]. A controlled study revealed something unexpected – trust in AI-generated content stayed similar across all age groups [13]. This suggests that age might not affect trust levels as much as we thought.
Tech-savvy and educated users
Education level helps predict who will use ai recipes. Studies show that people with higher education depend more on AI recommendations [14]. This goes beyond just knowing how to use technology. Education gives people both the technical knowledge to use AI tools and the critical thinking skills to evaluate what it all means when using ai generated recipes [14].
When we look at who actively uses AI, educated users show more willingness to try ai for recipes. This matches what we see with other types of technology – educated people tend to accept automated systems more readily [71, 72].
Food neophobia and openness to new experiences
Your personality, especially how open you are to trying new foods, shapes whether you’ll accept an ai recipe app. Being afraid of new foods (food neophobia) affects how people view and use this technology [15]. Children provide a good example. They usually fear new foods more than adults do, and they like AI-generated recipes less than those made by human chefs. The main reason? They see computer-created recipes as too unusual [1].
Trust builds through a mix of understanding and emotional connection. People start trusting ai recipes more as they learn about food technology, have good experiences with it, and develop positive feelings about using it [15].
The problem of naive AI realism
The way AI-generated food content looks creates a psychological trap. Researchers call this “naive AI realism“—people trust or doubt ai recipes too much because they look so real [1]. This bias affects how people review these recipes based on their real value.
Why realistic images can mislead
AI-generated food pictures next to new recipes look incredibly appetizing. These images score higher in likability than real food photos [1]. Studies show AI makes food look better by improving symmetry, color, and shine. These changes make food pictures more attractive [16]. Here’s an interesting finding: people found AI pictures more appealing than real food photos when they didn’t know AI created them [16].
AI food pictures show these common signs:
- Perfect symmetry and plating that’s too good to be true
- Surfaces that are too shiny and uniform
- Shadows or reflections that don’t match
- Food arrangements that defy physics
- More pictures of high-calorie foods [16]
Overtrusting AI based on presentation
This behavior is like traditional naive realism—the belief that our senses give us reliable information about our surroundings [17]. With ai generated recipes, people think good-looking food means cooking expertise. So realistic pictures make recipes seem more trustworthy, even when they have serious problems [17].
This creates problems with new ai recipe app suggestions. Human chefs might suggest creative but reliable changes. Similar ideas from ai in the kitchen systems might get dismissed as “hallucinations”—random outputs from language models [4]. DeepMind research shows tiny changes in AI pictures can affect human judgment, just like they affect AI models [16].
The gap between appearance and outcome
Naive AI realism becomes dangerous because pretty pictures don’t match real cooking results. BBC researchers proved this gap. They cooked AI-generated dishes next to chef-created ones. Tasters called the AI food “gray goop” and said it looked like “cat food,” despite the attractive preview images [5].
Untested ai recipes often skip vital details like measurements, cooking times, or specific techniques needed for success [18]. This gap between looks and results goes beyond disappointment. Missing safety warnings about food preservation or dangerous ingredient combinations can create real risks [18]. The false security from naive AI realism puts both cooking success and food safety at risk. Pretty pictures replace real cooking knowledge.
Designing AI recipe apps people can trust
“I think trust comes from transparency and control. You want to see the datasets that these models have been trained on. You want to see how this model has been built, what kind of biases it includes. That’s how you can trust the system. It’s really hard to trust something that you don’t understand.” — Aidan Gomez, Co-founder and CEO, Cohere; co-author of ‘Attention Is All You Need’
Modern ai recipe applications need architectural solutions that work to overcome basic trust barriers. Early systems focused mainly on generating outputs without caring much about reliability. Today’s platforms build user confidence through four essential design elements that ensure structural integrity.
Explainable AI and transparency
Trust in ai recipes starts with transparency. Today’s best practices use explainable AI (XAI) techniques. These techniques show clear reasons for ingredient choices and explain the logic behind changes step by step. 83% of respondents believe food companies should tell consumers when they use AI in their products [9]. This addresses the accountability gap found in consumer research. Modern ai recipe app systems now show visual hints to identify ingredients. This makes applications more user-friendly [19].
Context-aware suggestions
ai generated recipes now adapt to each user’s situation through smart filtering. Adapt-a-dish systems customize outputs based on dietary priorities, cooking skills, and portion sizes [20]. Chefmate AI changes basic keyword searches into conversation-style recommendations. It understands ingredient swaps, cooking methods, and diet limits [3]. This awareness helps ai in the kitchen applications give individual-specific guidance. The system considers both practical needs and the emotional side of cooking.
Confidence scores and safety warnings
Leading applications now use confidence scoring systems to fight blind trust in AI. These scores help users tell strong recommendations from uncertain ones [21]. Specialized tools like spice level detection show how sure the model is about suggested spice amounts [22]. Safety-focused designs also run automatic checks. They stop the creation of dangerous recipes by blocking toxic ingredient combinations [23].
Retrieval-augmented generation (RAG) for reliability
RAG technology stands as the most important advance in ai for recipes. It connects AI that creates recipes with trusted knowledge bases [24]. Unlike basic language models trained on general data, RAG systems find relevant recipes through meaning-based understanding [3]. This setup checks recipes against proven cooking knowledge in real time. It greatly reduces AI mistakes [25]. RAG improves reliability by linking recipe suggestions to factual, trusted sources. This fixes core trust issues without needing to retrain the whole model [26].
Conclusion
The relationship between AI and cooking shows a complex psychological pattern that we shaped through trust. AI recipe generators keep advancing, but people don’t deal very well with them because of deep psychological barriers. Trust isn’t just a technical challenge – it’s shaped by how accurate people think the AI is, their views on technology, and when they use it.
Studies show that trust builds in predictable ways. People first think it through logically before they might accept it emotionally. This works just like human relationships, where doubt slowly fades after good experiences. People are nowhere near as resistant to AI-generated traditional recipes compared to new creations. This shows how matching existing cooking norms makes acceptance easier.
The situation really affects trust decisions. People accept AI suggestions more readily for simple tasks like using leftovers or making basic meals. But when it comes to important occasions like dinner parties or cultural celebrations, they become much more skeptical. This shows how people weigh the benefits against possible AI mistakes.
Education and comfort with technology are the best predictors of who will use AI recipes. Naive AI realism is the biggest problem – when attractive presentations create false confidence and lead to a dangerous gap between what people expect and what they get.
Future progress depends on solutions that tackle these trust issues head-on. Explainable AI, smart filtering, confidence measurements, and improved generation methods all help make systems more reliable and transparent. These changes suggest that AI will gain more acceptance as it proves its accuracy while respecting cooking’s emotional side.
Notwithstanding that, AI’s future in cooking needs to balance tech capabilities with human needs. While AI offers clear benefits in efficiency and creativity, people will only accept it when systems respect cooking’s cultural importance and maintain safety standards. Trust isn’t automatic – it must be earned. That’s the key ingredient that connects what’s technically possible with what people will accept in this changing relationship between AI and cooking.
Key Takeaways
Understanding the psychology behind AI recipe trust reveals critical insights for both users and developers navigating this emerging culinary landscape.
• Context determines trust: People readily accept AI recipes for low-stakes cooking like using leftovers but avoid them for high-stakes scenarios like dinner parties or special occasions.
• Traditional recipes build confidence: AI-generated versions of familiar dishes like lasagna receive similar trust as human-created recipes, while innovative AI creations face significant skepticism.
• Visual appeal creates dangerous overconfidence: AI-generated food images often look more appealing than real food photos, leading to “naive AI realism” where attractive presentations mask recipe flaws.
• Demographics predict adoption patterns: Younger, educated, tech-savvy users with openness to new experiences show highest AI recipe acceptance rates across all studies.
• Trust requires transparency and safety measures: Successful AI recipe apps must incorporate explainable AI, confidence scores, safety warnings, and retrieval-augmented generation to earn user confidence.
The key to widespread AI recipe adoption lies not in perfect algorithms, but in building systems that respect both the functional and emotional aspects of cooking while maintaining transparency about their limitations.
FAQs
Q1. How do people’s trust levels in AI recipes vary based on the cooking context? People tend to trust AI recipes more for low-stakes scenarios like using leftovers or everyday meal preparation. However, trust decreases significantly for high-stakes situations such as dinner parties or culturally significant meals, where the perceived cost of potential AI errors is higher.
Q2. Why are traditional AI-generated recipes more trusted than innovative ones? Traditional AI-generated recipes align closely with users’ existing culinary experiences and expectations, triggering familiar mental schemas. In contrast, innovative AI recipes face higher skepticism due to unfamiliarity and concerns about AI’s ability to understand cultural nuances and flavor principles.
Q3. What is “naive AI realism” and why is it problematic for AI recipes? Naive AI realism refers to the tendency to overtrust AI-generated recipes based on visually appealing presentations. This is problematic because attractive AI-generated food images can create false confidence in recipes that may have serious flaws or safety issues, leading to disappointing or potentially dangerous outcomes.
Q4. Who is most likely to trust and use AI-generated recipes? Younger, educated, and tech-savvy individuals who are open to new experiences show the highest acceptance rates for AI recipes. These demographics typically have greater familiarity with technology and are more willing to experiment with AI-powered culinary tools.
Q5. How are AI recipe applications being designed to build user trust? To build trust, modern AI recipe applications incorporate features like explainable AI for transparency, context-aware suggestions, confidence scores, safety warnings, and retrieval-augmented generation (RAG). These elements aim to improve reliability, provide personalized guidance, and anchor recipe suggestions in authoritative culinary knowledge.
References
[1] – https://www.tandfonline.com/doi/full/10.1080/10447318.2025.2505154
[2] – https://osf.io/dzfbn_v1/
[3] – https://github.com/ThakkarVidhi/chefmate-ai
[4] – https://ora.ox.ac.uk/objects/uuid:2930b7a6-906f-4d08-a299-5c60799ec07a/files/r2b88qd31s
[5] – https://www.bbc.com/future/article/20240823-artificial-intelligence-wants-to-tell-you-what-to-cook-but-are-the-results-edible
[6] – https://www.sciencedirect.com/science/article/abs/pii/S1878450X24001069
[7] – https://independentaustralia.net/life/life-display/ai-recipe-generators-a-recipe-for-disaster,17946
[8] – https://www.npr.org/2024/09/23/g-s1-23843/artificial-intelligence-recipes-food-cooking-apple
[9] – https://abiggerconversation.org/ai-designed-food-a-recipe-for-distrust/
[10] – https://theintelligentfriend.substack.com/p/can-we-accept-an-ai-chef
[11] – https://www.askattest.com/blog/research/ai-in-the-kitchen-the-future-of-food-or-recipe-for-disaster
[12] – https://www.askattest.com/blog/articles/2025-consumer-adoption-of-ai-report
[13] – https://studenttheses.uu.nl/bitstream/handle/20.500.12932/45178/Master Thesis – Sophie Heezen – Final.pdf?sequence=1&isAllowed=y
[14] – https://www.researchgate.net/publication/383747966_The_Impact_of_Education_Level_on_AI_Reliance_Habit_Formation_and_Usage
[15] – https://pmc.ncbi.nlm.nih.gov/articles/PMC11431241/
[16] – https://www.acquainted.studio/content/ai-generated-food-images-look-tastier-than-real-ones
[17] – https://www.researchgate.net/publication/385106997_Appealing_but_misleading_a_warning_against_a_naive_AI_realism
[18] – https://www.fodmapeveryday.com/how-to-detect-ai-generated-recipes-and-images-online/
[19] – https://medium.com/design-bootcamp/case-study-recipe-app-e12ee8f22ba6
[20] – https://github.com/Atul-AI08/Context-Aware-Recipe-Generator
[21] – https://www.linkedin.com/pulse/from-ingredients-insights-lessons-building-ai-recipe-rater-lobo-0l0se
[22] – https://www.nyckel.com/pretrained-classifiers/spice-levels/
[23] – https://www.opb.org/article/2024/09/23/ai-can-generate-recipes-that-can-be-deadly-food-bloggers-are-not-happy/
[24] – https://blogs.nvidia.com/blog/what-is-retrieval-augmented-generation/
[25] – https://arxiv.org/html/2411.08715v1
[26] – https://www.ibm.com/think/topics/retrieval-augmented-generation
