Leveraging a team of LLMs
Now you can leverage multiple LLMs, for free, to help you get the best genAI deliverable while you experiment and choose your preferred infrastructure.
HuggingChat gives you a chat-based interface AND the ability to select from a set of different LLMs. You can run at least a few queries for free (I didn't hit a limit yet).
I wanted to see how they compared, and at the same time understand how well LLMs might help to implement a content strategy for the upcoming redesign of our 55BirchStreet website. So I chose a hypothetical example of a blog post about genAI and sustainability as a concrete example.
When I'd got a set of 5 responses from HuggingChat, I fed them all into ChatGPT4, and asked it to analyse and evaluate them. It then suggested a hybrid version would be the most effective, so I got it to write a definitive version.
Below is the brief, the analysis and the final version. (You can hit me up on LinkedIn if you’d like to see all the responses in full.)
I don’t even know how much we will really use LLMs in content production, but the ability to generate multiple responses from different LLMs, and have them evaluated by another LLM with a final ‘best in class’ example, is a really interesting capability.
Overall brief
Based on a keyword/phrase, write a blog post for a specific website.
Specific context
This keyword "How does AI contribute to sustainable development goals?"
For a blog post on this website www.55birchstreet.com
Guidance for final deliverable
Write it in a way that makes it relevant and inspiring for any business involved in making physical goods, with a global supply chain.
The goal is to inspire them to learn more. The CTA is for them to contact 55BirchStreet.
Models used
Response 1: Hugging Face HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1
Response 2: Google google/gemma-1.1-7b-it
Response 3: Mistral mistralai/Mixtral-8x7B-Instruct-v0.1
Response 4: Cohere CohereForAI/c4ai-command-r-plus
Response 5: Meta meta-llama/Meta-Llama-3-70B-Instruct
ChatGPT-4’s analysis of the LLMs responses
Overview:
Each response addresses the given brief, offering valuable insights into how generative AI can contribute to sustainability efforts in global supply chains. They all follow a similar structure, including an introduction, analysis of relevant AI applications, and a call to action urging readers to contact 55BirchStreet. Despite the shared theme, each blog has distinct strengths and opportunities for improvement.
Response 1: Hugging Face (HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1)
• Strengths: This post is structured with clear, concise sections that efficiently describe AI's specific contributions to supply chain optimization, sourcing, quality control, traceability, and circular economy. The language is persuasive yet approachable, which suits the intended audience.
• Weaknesses: The writing sometimes feels formulaic, lacking original insights or compelling storytelling that would make the post more memorable. It could also expand on real-world examples of AI in action to offer deeper practical insights.
Response 2: Google (google/gemma-1.1-7b-it)
• Strengths: This post successfully balances clarity and inspiration with comprehensive overviews of sustainability challenges and AI solutions. The suggestions for smart transportation, predictive modeling, and automated manufacturing are practical, and the call to action clearly directs readers to explore further.
• Weaknesses: It would benefit from providing more concrete examples, especially for the challenges in sourcing sustainable materials. The post is also somewhat redundant in parts, repeating similar ideas without adding new value.
Response 3: Mistral (mistralai/Mixtral-8x7B-Instruct-v0.1)
• Strengths: This post dives deeply into specific challenges and how AI can directly alleviate them, particularly around compliance, quality control, and responsible sourcing. The detailed descriptions and varied suggestions for AI-driven innovation make it highly informative.
• Weaknesses: The overall writing could be more engaging, as it reads heavily technical and lacks a broader strategic focus. Also, sections are dense, making it harder to follow, and practical steps to start implementing AI could be clearer.
Response 4: Cohere (CohereForAI/c4ai-command-r-plus)
• Strengths: This response integrates a comprehensive list of actionable insights on data-driven decision-making, logistics, and predictive maintenance. The post also highlights ethical and sustainable sourcing practices and balances these with a focus on achieving SDG goals.
• Weaknesses: It tends to cover a broad range of topics without diving deep into each one. It could also be more tailored to the target audience by incorporating examples specific to their manufacturing challenges
Response 5: Meta (meta-llama/Meta-Llama-3-70B-Instruct)
• Strengths: The post is highly persuasive, blending strategic insights with concrete examples like predictive maintenance and circular economy models. It compellingly lays out the business case for AI in supply chains while providing a logical roadmap for implementation.
• Weaknesses: The writing is less technical than the other posts, which could be a weakness if the target audience seeks in-depth analysis. It also lacks specific advice for overcoming practical challenges in sustainable sourcing or meeting compliance standards.
Overall Recommendation:
For the target audience of project managers, AI managers, and sustainability managers in companies with global supply chains, Response 5 (Meta) would likely be the most effective to deploy. It has a strong combination of persuasive strategic insights and practical suggestions while offering a comprehensive call to action. However, incorporating elements from the other responses—such as the specific technical details in Response 3 (Mistral) or the comprehensive ethical considerations in Response 4 (Cohere)—would enhance its value further.
Final ChatGPT-4 response
Title: Unlocking Sustainable Supply Chains with AI: How Businesses Can Make a Global Impact
Introduction:
In today's rapidly evolving business landscape, organizations face immense pressure to address environmental concerns while maintaining profitability. Manufacturing companies that operate global supply chains are at the forefront of this challenge. Fortunately, generative AI emerges as a transformative force, offering innovative solutions to optimize supply chain operations while driving positive environmental impact. This blog post explores the many ways AI can empower your business to meet the United Nations' Sustainable Development Goals (SDGs) and create a more sustainable future.
AI-Powered Solutions for Supply Chain Optimization:
Generative AI offers various tools and strategies for optimizing supply chains:
Predictive Modeling: Advanced AI algorithms can analyze historical and real-time data to predict future resource demand. This enables better production planning and inventory management, reducing waste and ensuring efficient resource utilization.
Smart Transportation: AI can help identify the most efficient logistics networks, reducing emissions and improving delivery times through optimized routing.
Sustainable Sourcing: AI-powered tools offer enhanced transparency into suppliers' social and environmental impact, helping you identify sustainable suppliers and ensuring ethical practices in raw material sourcing.
Fostering Sustainable Growth with Circular Economy Models:
AI can facilitate your transition to a circular economy by designing products for reuse, refurbishment, or recycling. By analyzing production data and customer feedback, AI algorithms can optimize reverse logistics, enabling products to be returned, refurbished, and resold efficiently. This minimizes waste and promotes a sustainable product lifecycle.
Ensuring Compliance and Responsible Sourcing:
The rapidly evolving landscape of global regulations makes compliance challenging, particularly in areas like labor rights and environmental standards. AI can help you monitor your suppliers' adherence to these regulations by automating the review of legal documents and supplier audits. This proactive approach to risk management strengthens your reputation and helps you maintain supply chain continuity.
AI-Driven Innovation and Data Insights:
Data has emerged as a crucial driver of modern business innovation. Generative AI excels at extracting actionable insights from vast data sets, enabling rapid experimentation and shorter go-to-market timelines. Computer vision-based collaborative robots ("cobots") transform assembly lines into adaptive workspaces, offering continuous improvement. Generative design algorithms can also identify lightweight yet durable parts optimized for both function and sustainability.
Ethical and Governance Considerations:
As with any new technology, AI implementation requires careful governance. It’s crucial to address challenges around data privacy, algorithmic bias, and potential job displacement. Responsible AI governance ensures transparency and accountability, with clearly defined metrics to monitor progress toward sustainability goals.
Conclusion:
Generative AI offers unparalleled opportunities to revolutionize global supply chains, helping you reduce waste, optimize resources, and promote sustainable sourcing. By leveraging its potential, your business can reduce costs, enhance brand reputation, and contribute to a more sustainable future.
Call to Action:
Ready to unlock the potential of AI for a more sustainable tomorrow? Contact us at 55BirchStreet to schedule a consultation and explore how generative AI can transform your global supply chain.
