ServiceNow Inc., a $220 billion company, held a conference in Toronto to discuss the potential of generative AI. CEO Bill McDermott highlighted the importance of AI in driving customer value and transforming the industry. 

The conference featured booths offering various ways to use AI, such as unlocking the future of work with AI agents, enhancing IT experiences, building no-code apps, simplifying procurement, and achieving next-level procurement. However, the reality of generative AI today is more esoteric and less glamorous than its hype.

Proponents of generative AI have promised efficiency gains, cost savings, and supercharged productivity for individuals, companies, and entire economies. However, these tasks are generally low-hanging fruit and do not always translate into significant cost savings or economic benefits. A study by Kyndryl found that only 41% of Canadian executives are seeing a positive return on their investments due to some companies not yet being ready from a technology and data perspective.

The second major obstacle is the price performance of AI. While the technology is powerful, it is not the expensive labour for automating rote work within the business. In conclusion, while generative AI may be coming into view, it is still out of reach for many companies.

Companies are increasingly using generative AI to streamline processes and save time. However, only 30% of executives view AI primarily as a revenue driver or cost reduction. Nonetheless, small changes can make a significant difference. 

Colligo Networks uses generative AI software for HR functions, while Odaia Intelligence helps pharmaceutical companies identify patient populations for new drug treatments. 

Boosted.ai’s software allows finance professionals to analyze hundreds or 1,000 times more data than they would in a normal environment. 

Neo Financial Technologies has deployed a generative AI customer service chatbot, handling 50% to 60% of customer problems without escalating to a human agent. 

Toronto-Dominion Bank is deploying a chatbot to its 1,200 call centre workers, which has led to a 15% reduction in hold times and a 15% drop in the number of times customer service reps have to speak to their managers. The bot provides a link to source material, which call centre reps are supposed to verify. Coveo Solutions Inc. is using generative AI technology to power help sections on websites for companies like 

Dell Technologies and United Airlines, reducing customer service case loads by 31% and seeing a 20% boost in customer resolution within the first six weeks.

The economic impact of generative AI is uncertain, with estimates suggesting it could add $187 billion annually to the Canadian economy by 2030, pushing productivity growth from 0.6% to an “astounding” 8% if quickly implemented. Microsoft Corp. estimates it could add $230 billion to the economy and save the average Canadian worker over 175 hours a year. However, economist Daron Acemoglu estimates the technology will only increase productivity in the United States by 0.5% cumulatively in the next decade due to its inability to handle complex problems and limited menial labour.

Avi Goldfarb, a University of Toronto professor and Rotman Chair in Artificial Intelligence and Healthcare, believes AI fits the bill, as it is being adopted across industries and carries the potential to spur further innovation. However, the Dais think tank at Toronto Metropolitan University found no strong effect on short-term productivity, and the companies that adopted AI were already more productive than their peers.

The costs of generative AI are increasing, with training advanced large language models (LLMs) becoming more expensive due to the number of graphics processing units needed, time involved, and post-training tinkering required. The largest models are expected to cost more than $1 billion to train by 2027. Inference, the process of accessing LLMs, is also becoming more expensive for customers. 

The economics of inference didn’t matter much until a couple of years ago, but the recent surge of interest has compelled companies to find ways to improve efficiency and lower costs. LLM developers like OpenAI and Anthropic are struggling to turn a profit, but Nick Frosst, co-founder of Cohere Inc., believes that generative AI will become more useful to businesses with the right tweaks and customizations. 

Source: Globe and Mail