Sergey Diakov
⬤ AI systems are improving at rates that can be measured and tracked, not just imagined in sci-fi scenarios. Performance benchmarks show steady climbs across multiple categories, with some areas experiencing exponential growth. These aren't wild predictions—they're based on actual data points collected as large-scale AI systems continue evolving.
⬤ The most interesting developments are happening in planning and reasoning, where AI capabilities have been climbing consistently. If you plot the performance curves and extend them forward, the math suggests these systems could hit human-level benchmarks in roughly five years. That's not a promise, just what the numbers indicate when you follow the trend lines. It's the same kind of analysis used in any field trying to understand where technology might be headed next.
⬤ Of course, nothing's guaranteed. The upward momentum is real and measurable, but technical roadblocks could slow things down or change the trajectory entirely. Still, looking at empirical data and extrapolating forward isn't guesswork—it's how you make informed assessments about future possibilities. The trends are strong enough that ignoring them wouldn't make sense.
⬤ What makes this discussion different is the focus on actual evidence rather than hypothetical what-ifs. If planning and reasoning benchmarks keep climbing at current rates, it'll reshape expectations about how quickly AI becomes truly capable. But the built-in uncertainty keeps things grounded—these projections depend on trends continuing, which means there's room for both optimism about breakthroughs and realism about limitations.
Sergey Diakov
