Is AI Everything?

Notes from GITEX Global 2023, Dubai

Last week was a hectic but enjoyable blur of visiting GITEX Global 2023 in Dubai to see hundreds of technology companies and meet practitioners from all over the world. The theme this year was “AI Everything” which fit neatly into my current focus.

A few notes and observations:

1) Artificial Intelligence (AI) has really caught the imagination of global enterprises, investors and entrepreneurs. The number of companies showcasing their capabilities with AI-based solutions was awe-inspiring.

(a)??LLM models with Generative AI capabilities are powering a lot of new realistic avatar-based natural language capable chatbots

(b)??Many experienced hands, such as Sanchit Suneja , Product Director at Booking.com, have already been heavily using AI in their core platforms. Booking.com seems to have 350 ML models in production which deliver 200 billion predictions per day! Most of this pre-dates the emergence of Generative AI (GAI)

(c)??Picks-and-shovels for AI related use cases are not only being provided by the large technology players like Microsoft, Google, AWS, Adobe, Dell, Snowflake etc., but newer businesses are emerging who want to provide developers and non-tech users an easy way to build AI-based apps. Such development platform startups are gathering breathtaking amounts in funding.

• Prag Sharma , Global Head of AI at Citi, warned that anyone developing software without using the already available AI tools like GitHub Copilot will soon find themselves abandoned by talented developers

(d)???AI is not just a tech phenomenon but a business phenomenon. Business leaders and advisors are fully alive to the possibilities of AI. Some examples:

• Carolina Pinart, Ph.D. , Group Head of R&D IT at Nestle, spoke of several ways in which Nestle uses AI for new offering development (coffee and cheese, anyone?) to customer segmentation to identifying efficiency gains.
• Prag Sharma , Global Head of AI at Citi, spoke about using AI in core areas like client engagement, operational enablement and risk & compliance.
• Vilmos Lorincz , MD Data Product at Lloyds Banking Group, had interesting examples of using AI to predict current account balances and to launch a new service for bereavement events.

(e) The pursuit of AI is gender-neutral – this was certainly not your typical cybersecurity or gaming conclave! The GITEX organizers must be congratulated to have found and convinced so many experienced women who are experts in this field to travel and freely share their knowledge. For example,

• Sol Rashidi, MBA , CDO/CAO at Estée Lauder, with multi-decadal experience of having delivered AI use cases and having fought the good fight and surviving to warn us of the common mistakes made around not securing budgets beyond POC, not investing in the right infrastructure, or communicating about AI with your Board in language beyond the primary school level!
• Franziska Bell, PhD , SVP Digital Technology at BP, spoke about the prerequisites for unlocking the potential of AI, which include developing a collaborative ecosystem, ensuring compliance and security, and developing in-house AI expertise.
• Saskia Steinacker , Global Head of Strategy & Digital Transformation at Bayer, spoke eloquently about the use of AI in drug development and other pharma products.

2)??At the same time, many experienced business leaders were consistent in their sobering assessment that this is the moment of maximum hype and a time to be careful of it:

• Dr. Luc JULIA , the Chief Scientific Officer at Renault, showed a Gartner chart with AI right at the top of their Hype Cycle…which ominously turns down precipitously from here.
• Prag Sharma of Citi insisted that “boring AI” is where the focus should be and where real use cases currently reside
• Dr. Akram Awad , Partner at BCG and Awad El-Sidiq , Head of AI & Analytics at ADNOC stressed on the importance of a coherent data strategy before building the AI superstructure on it
• Daniel Wu , Head of AI & ML at J.P. Morgan Chase, suggested that Generative AI is in its early days and there is a lot more work to be done.
• Shameem Farouk , EVP and Head of Digital Capabilities at Maybank, shared their quest for humanising financial services and using rule-based engines which may be safer than GAI for customer interfacing.

Every single person talking about GAI also spoke about challenges with hallucination, inaccuracy and lack of explainability. The experts are in violent agreement that GAI is good at some jobs but not good at many others.

3)?Benefits may be imaginary but costs are real:

• Alex de Vries-Gao , founder of Digiconomist, shared his freshly published research which says that AI use cases can consume enough electricity to power a small city! His research indicates that the cost of using AI-powered Google search could use 7x the energy used in a standard non-AI search!
• A data scientist developing LLMs for healthcare use cases told me that they estimated a one-time cost of $30m for model setup and training, and then $2m/day to run a ChatGPT based service at their scale.

In summary, I came back from GITEX hugely inspired to see the developments happening at the speed of AI and the variety of uses it is being put to. But many questions remain to be answered around the costs of using Generative AI at scale and the pitfalls to avoid.

** Any mis-quotes above are inadvertent. Please accept my apologies and tell me so I can correct them.