Street Cred: Making GenAI work for your Business and Teams 
Creator: encrier | Credit: Getty Images/iStockphoto Copyright: encrier

Street Cred: Making GenAI work for your Business and Teams 

Following the exciting BigDataLDN event last week I wanted to share a shortened version of my talk here. I'd be very interested in your thoughts, use cases and experiences with LLMs - please comment below; Also, if you are interested in getting the inside track and are willing to participate in our experiments in this area - send me a DM.

The Aviation Paradigm: ✈️ Imagine a world where generative AI not only transforms regular businesses but also reshapes the aviation landscape, assisting in better aviation decision-making, from flight, fleet and network planning to complex financial risk management and seat level carbon emissions. But how do we ensure such AI systems are trustworthy, reliable, and grounded in realism? This domain, where safety is paramount and human's trust in data is the backbone, offers a unique lens to view the challenges and promises of GenAI.

RELX and Responsible AI: 🌟 It is with these challenges in mind that Cirium, as part of RELX, updated our public commitment in 2022 to responsible AI with principles that resonate deeply with the current focus on LLMs and Generative AI. Here's a glimpse of the RELX Responsible AI Principles

  1. Prioritizing the real-world impact of solutions on people. 

  2. Actively preventing the creation or reinforcement of unfair bias. 

  3. Ensuring explainability in how solutions operate. 

  4. Establishing accountability through human oversight. 

  5. Upholding privacy and championing robust data governance. 

But what do these principles mean in practice? Do we actually want an assistant in the room, in the slack channel, in the Teams meeting, who is indistinguishable from a human colleague?

Deconstructing the Desired UX with LLMs: 🔍 The experience of interacting with LLMs, especially ones as sophisticated as GenAI, demands careful thought: 

  • 1️⃣ Human-centric Interaction: Should LLMs simulate human-like communication? A model could be constructed that is steeped in human nuances, following team specific shorthand, using internal acronyms including humor or colloquialisms, and organic interactions. Yet, this path might also unveil challenges in bias and ethical realms. Should the LLM adopt a neutral view or adopt a highly risk averse role at the expense of positive scenario exploration?

  • 2️⃣ Unambiguous AI Communication: Or, should LLMs be explicitly machine-like, emphasizing clarity over mimicry? This ensures information is relayed with precision, sidestepping potential ambiguities. You always know "who" is talking.

Balancing Authenticity with Clarity: ⚖️ The equilibrium between human mimicry and lucid communication rests on: 

  • 1️⃣ Purpose-driven Design: Shape your LLM strategy around distinct low stakes aviation "business" use-cases (nothing safety critical) - there are huge areas for knowledge worker productivity improvements in aviation, for example, just summarising complex documentation, uncovering anomalies and conflicting requirements. While customer interactions might benefit from a human touch, technical tasks demand unambiguous clarity - including tone of voice.

  • 2️⃣ Audience Insight: Recognize your user base and model your archetypes. If they seek interactions reminiscent of human conversations, then tuning LLMs in that direction amplifies user delight and enables a more seamless exchange - using customer shorthand specific to a team can build stronger trust. 

  • 3️⃣ Treading the Ethical Line: Contemplate the ethical dimensions of crafting AI that's indistinguishable from humans. A creative approach is often required for business ideas and product pitches - these require a less precise argument and benefit from creative chains of thought.

In Closing: 🚀 As we integrate LLMs into business processes and low stakes aviation decision making it's crucial to strike a chord between human authenticity and crystal-clear communication. 

Bear in mind, there's no one-size-fits-all. The nuances of your organisation dictate the tuning of GenAI. So, grasp these UX insights, align them with your goals, and watch GenAI elevate your business landscape.

#genAI #ArtificialIntelligence #BusinessRevolution #RELX 

 

David Alexander Eristavi

Customising AI to your needs | The AI & Data specialists

1y

Aviation Festival, was such an great time!! Great read... who wouldn't want an assistant acing the Voight-Kampff test?

Like
Reply

To view or add a comment, sign in

Insights from the community

Others also viewed

Explore topics