The Challenges of Deploying Generative AI in Production

5min Snip

00:00

Play full episode

Summary

Transcript

Episode notes

The challenges associated with deploying generative AI in production. The two things maybe that I hear the most from industry is one around that concept of hallucination. We still have to create an inclusive data set, even though the data set in GPT is much or generativeAI in general, much smaller. There's lots of techniques now to control how a response is generated. And we do need to take more time, more care and more effort in training them.

With the advent of any new technology that promises to make humans lives easier, replacing concious actions with automation, there is always backlash. People are often aware of the displacement of jobs, and often, it is viewed in a negative light. But how do we try to change the collective understanding to one of hope and excitement? What use cases can be shared that will change the opinion of those that are weary of AI?

Noelle Silver Russell is the Global AI Solutions & Generative AI & LLM Industry Lead at Accenture, responsible for enterprise-scale industry playbooks for generative AI and LLMs. In this episode of our AI series, Noelle discusses how to prioritize ChatGPT use cases by focusing on the different aspects of value creation that GPT models can bring to individuals and organizations. She addresses common misconceptions surrounding ChatGPT and AI in general, emphasizing the importance of understanding their potential benefits and selecting use cases that maximize positive impact, foster innovation, and contribute to job creation.

Noelle draws parallels between the fast-moving AI projects today and the launch of Amazon Alexa, which she worked on, and points out that many of the discussions being raised today were also talked about 10 years ago. She discusses how companies can now use AI to focus both on business efficiencies and customer experience, no longer having to settle for a trade-off between the two.

Noelle explains the best way for companies to approach adding GPT tools into their processes, which focusses on taking a holistic view to implementation. She also recommends use-cases for companies that are just beginning to use AI, as well as the challenges they might face when deploying models into production, and how they can mitigate them.

On the topic of the displacement of jobs, Noelle draws parallels from when Alexa was launched, and how it faced similar criticisms, digging into the fear that people have around new technology, which could be transformed into enthusiasm. Noelle suggests that there is a burden on leadership within organizations to create a culture where people are excited to use AI tools, rather than feeling threatened by them.