The panel, moderated by George Roukas, co-founder and former partner of Hudson Crossing, also included:
- Mark McSpadden, VP of Product at Amex GBT
- Ed Silver, co-founder of Travel Again and former CIO at iSeatz
- Dmitry Baikov, Technical Director of AI/ML at DataArt
The objectives of the webinar were to:
- Show that generative AI is real and ready to use today
- Explore proofs of concept for revenue generation and cost savings
- Encourage participants to try generative AI themselves
During the live webinar, experts covered the following points:
- Updates on the state of generative AI in travel
- POCs reviewing
- Revenue generation and cost-saving opportunities
- Cautions and considerations in using generative AI
Generative AI Solutions Overview
Dmitry Baikov, Technical Director of AI/ML at DataArt, demonstrated the use of some practical applications of generative AI in the travel sector and showcased the speed and versatility of generative AI in real-world proof-of-concept examples, such as email detection and text extraction, email follow-up generation, and support automation for helpdesk teams. Dmitry emphasized the ability of generative AI to enhance productivity and enable support staff to concentrate on more intricate tasks. Exploring the possibilities further, he showcased how chatbots could be built on existing information indexed in confluence or databases, facilitating efficient data validation, mining, and mapping.
Potential Risks of AI for Businesses
Ed Silver, co-founder of Travel Again, highlighted the cost reduction and speed enhancements that can be achieved through AI, such as breaking down emails and creating pre-responses for agents. He encouraged everyone, regardless of technical expertise, to explore automating their own business workflows and test the capabilities of generative AI.
However, he also expressed caution and skepticism about stretching AI beyond its current usefulness. He stressed the importance of solving real customer problems and ensuring that AI implementations add genuine value to the user experience.
How to Optimize Customer Service with AI
Dmitry provided insights on three approaches to achieving customization with generative AI:
- Prompt engineering: Companies can shape AI responses by providing specific examples and guidance. This includes using phrases, data, and guidelines that reflect the company's rules, terms, and conditions.
- Indexing approach: This method involves splitting data into smaller pieces and storing them in a vector database. By extracting the context from these pieces of data and comparing it to specific questions, companies can retrieve relevant information from different sources. This approach expands the context size and enables more comprehensive responses.
- Fine-tuning: Recommended for larger companies with abundant data, fine-tuning involves refining the AI model to align with specific requirements and objectives. By leveraging company and customer data, the model can be trained to understand better and respond to unique needs.
The panelist emphasized the importance of starting with quick wins and gradually advancing AI capabilities, highlighting the value of using pre-trained models and making adjustments based on specific business needs.
Q&A Session
In this part of the webinar, the panelists answered some questions from the audience. Here is a summary of the Q&A session:
Question: Are agents ready to incorporate reply generation into their workflow?
Agents can benefit from using reply generation technologies in their workflow, as seen in successful pilots with small groups of agents and customers. It allows them to have a starting point, quality check the information, and add a personalized touch to the responses.
Question: Will you still be competitive 12 months from now if your competitors have already started using this tool?
It depends on the product, customer base, and industry. Being 12 months behind gives competitors a significant advantage, so there should be a sense of urgency to test and learn how generative AI applies to customer problems and business. Starting early helps gain specific insights and adapt to behavioral changes in interactions.
Question: How does the panel see the role of automated AI bots interacting with generative AI to change the travel search and evaluation process?
Consumers may prefer chat-based search experiences in the future, but it's still being explored. Generative AI can accelerate personalization by understanding user preferences through chat interactions. Combining generative AI with traditional UI and interconnecting interfaces holds potential for the future.
Question: What kinds of jobs will be available in the future, and what kinds are likely to be automated away?
Automation can handle simple tasks, allowing travel counselors to focus on more complex and high-value aspects. Automation use cases remain viable for travel counselors. Expanding expertise in utilizing AI tools for business workflow improvement creates new job opportunities.
Question: Can you talk about the implementation time of a solution for email and text extraction and reply generation?
Implementation time for a proof-of-concept (POC) varies based on specific integration goals. Platforms like ChatGPT provide initial results quickly. Typically, it takes 4 to 8 weeks at DataArt to deliver a POC in generative AI or machine learning.
Question: How can we help streamline operational airline schedules, flight, and pilot scheduling?
Machine learning models and mathematical approaches are more suited for scheduling complexities. However, exploring opportunities for innovation in airline operations, particularly in communication aspects, can be valuable. Generative AI models like ChatGPT may not be the primary solution for scheduling tasks at this time.
Conclusion
To delve deeper into this webinar, I invite you to watch the full recording here. Explore the transformative potential of generative AI in travel, gain insights from industry leaders, and discover how AI can revolutionize customer experiences and drive business success.
The content and insights shared in this webinar can empower you to leverage generative AI in your own travel business, so be sure to watch the full session for a comprehensive understanding. And if you have any questions, please contact us: aiml@dataart.com.
