Jason’s scenario implied that if AI customer-support systems improved quickly enough, they could actually be handling ~98% of customer queries by the end of 2024. In reality, available data from 2024–2025 shows automation and AI assistance rising fast, but nowhere near 98% of queries being fully handled by AI.

For example, Comm100’s 2025 Live Chat Benchmark (based on 220M+ interactions) reports that AI chatbots handled about 73.8% of live-chat interactions in 2024, not 98%. More importantly, only about 45–46% of chats were fully resolved by chatbots alone in both 2023 and 2024, meaning most conversations still required a human agent to finish the job. (comm100.com)

Industry-wide summaries likewise talk about 60–85% of customer interactions being handled or assisted by AI/automation by 2024, and projections of ~75% handled by AI by 2025—again, well below 98% complete automation. (gitnux.org) Even leading-edge deployments aren’t there: Reuters reports India-based LimeChat’s bots currently handle about 70% of customer complaints for clients and aim for 90–95% within a year, i.e., still short of 98% and only for particular customers, not the whole market. (reuters.com)

User-experience studies also show that many customers still feel they need humans to get issues properly resolved: a 2024 UK survey found 54% of users say their problems are only truly solved by speaking with a human, and 40% would pay extra to avoid dealing with AI chatbots. (techradar.com) This is inconsistent with a world where ~98–99% of queries are fully resolved by AI.

The second part of Jason’s statement (reaching 98–99% in about four years if improvement is only ~10% per year, i.e., around 2028) concerns a future date and can’t yet be directly tested. But the concrete, time-bound claim that AI would be doing about 98% of queries by the end of 2024 clearly did not happen. Therefore, overall this prediction is best classified as wrong.