Generative AI is transforming content creation, marketing, and automation. This study develops an AI-powered Telegram bot integrating OpenAI’s DALL·E 3 and GPT-3.5 to generate marketing images and promotional texts efficiently. By leveraging AI, the bot reduces manual effort while maintaining creativity.
The methodology involves integrating OpenAI’s APIs for image and text generation, processing user inputs, and handling errors. Experiments assessed content quality, engagement, response time, and adaptability.
Results show that AI-generated content meets advertising standards, ensuring clarity and relevance. The bot efficiently handles diverse inputs, demonstrating scalability in automated marketing.
In the digital era, content marketing has become a key strategy for businesses to attract audience attention. One of the most advanced technologies used in advertising today is Generative AI, which enables the automatic generation of visual and textual content. With a Telegram bot powered by Generative AI, companies and individuals can instantly create advertising images and captions, eliminating the need for high-level design or copywriting skills.
According to data from Statista and Bloomberg, the adoption of Generative AI in marketing continues to rise (Figure 1). This technology is being leveraged to create dynamic ads, personalized campaigns, and enhance audience engagement. Moreover, market projections indicate that revenues from Generative AI are expected to surpass $1,500 billion by 2032, highlighting its crucial role in the creative industry (Figure 2).
While Telegram bots powered by Generative AI offer an innovative solution for advertising content creation, several key challenges remain:
This research aims to:
One of the main challenges in generating content using an AI chatbot is ensuring that the model correctly interprets the given prompt to create the desired image. AI models often experience hallucinations, where the generated output does not align with user expectations. This issue arises due to the model's varying interpretations of textual prompts, making it crucial to refine prompt engineering strategies to enhance output accuracy.
To develop a generative AI chatbot for advertising content, the following technologies and services are utilized:
The AI chatbot implementation follows a structured approach with three main steps:
1.Effective Prompt Design
Developing prompt templates that guide the AI to produce high-quality and relevant content.
2. API Integration
Connecting OpenAI’s API with the Telegram API to enable real-time input processing and content generation.
3. Cloud Deployment
Hosting the chatbot on PythonAnywhere to ensure accessibility at all times.
Despite the advantages of AI-generated content, there are some limitations to consider:
The AI chatbot offers distinct advantages over traditional content creation methods:
To ensure smooth operation, the following prerequisites are necessary:
The project's source code and documentation are available in a Github Repo, allowing developers to review and modify the implementation as needed with MIT License.
To install and use the chatbot, please refer to the GitHub repository below:
🔗 Github Repo
Run the following commands to clone and set up the project:
git clone https://github.com/Supra-San/GenAI-TeleBot.git cd GenAI-TeleBot. pip install -r requirements.txt
Set Up API Keys Edit config/config.py and replace placeholders with actual API keys.
OPENAI_API_KEY = "your_openai_api_key" TELEGRAM_API_KEY = "your_telegram_api_key"
Run the Bot
python bot.py
For this part, we will showcase screenshots or workflow diagrams that illustrate how the AI-powered Telegram bot generates advertising content. The key aspects to highlight include:
Hallucination in AI image generation refers to instances where the model produces visual elements that do not match reality or deviate from what was requested in the prompt. This often happens when the prompt is too vague, unspecific, or ambiguous. When a prompt lacks clarity, the model may "invent" details that were not specified, leading to images that do not meet expectations. This can be a major issue in advertising, where visuals must align precisely with the intended marketing message.
Examples of Bad vs. Good Prompts in Advertising Image Generation
🔴 Bad Prompt (Ambiguous & Unspecific)
📝 "Popcorn ad with 10% discount."
👉 Issues:
Even if the image is successfully generated, small details on the popcorn bucket may contain random, gibberish text instead of proper branding or promotional messages, a common hallucination issue in AI image generation.
✅ Good Prompt (Specific & Detailed)
📝 "A high-resolution advertisement of a large bucket of golden, buttery popcorn overflowing, placed on a modern movie theater counter with a bright neon sign in the background. The sign displays ‘10% OFF’ in bold, glowing red letters, with a tagline ‘This Week Only!’ underneath in a stylish font. The scene is set in the evening, with warm ambient lighting reflecting off the popcorn bucket. The advertisement follows a modern, cinematic aesthetic with a slight glossy texture, ensuring an eye-catching and appetizing presentation. The perspective is a close-up shot, highlighting the texture of the popcorn and the promotional text clearly visible."
👉 Advantages:
The evaluation of the AI bot focused on three main aspects: content quality and relevance, efficiency compared to traditional methods, and Optimization Integration.
| Content Creation Stage | AI-Assisted (Minutes) | Manual (Minutes) | Notes |
|---|---|---|---|
| Research | 5 – 25 | 5 – 25 | AI is not used for this stage in bots |
| Design & Visuals | 3 | 10 – 30 | Manual design requires more time for customization |
| Copywriting | 3 | 5 – 20 | AI can generate text swiftly |
| Engagement & Optimization | 5 | 5 – 15 | Both methods require time for audience interaction and post optimization. |
| Total Time | 16 - 36 Minutes | 25 – 90 Minutes | AI significantly reduces content creation time |
Integration: The bot’s seamless integration with Telegram provides users with an easily accessible platform for content creation. The user-friendly interface facilitates prompt input and content retrieval, simplifying the entire process.
Limitations: Despite its advantages, AI models sometimes produce output that requires human supervision, especially in maintaining brand voice and visual identity. Proper prompt writing is essential to mitigate this issue.
To effectively maintain and support the bot:
The implementation of AI in advertising content creation shows significant efficiency compared to manual methods. With a creation time of only 43 minutes compared to 135 minutes manually, AI allows businesses to produce content faster and be responsive to market trends.
Although AI speeds up the design and copywriting process, human validation is still needed to ensure quality and alignment with brand identity. Integrating AI into creative workflows can save time and resources, but optimizing prompts and revisions remains a key factor in achieving the best results.
There are no models linked
There are no datasets linked
There are no datasets linked
There are no models linked
