Why Today's AI Stops Learning the Moment You Hit "Deploy"
forbes.com
forbes.com
In any dynamic environment—robotics, autonomous agents, healthcare—this rigidity seems like a fundamental flaw.
2. Is fine-tuning doing more harm than good in real-world AI? “Fine-tuning a model is less resource-intensive than pretraining it from scratch, but it is still complex, time-consuming and expensive, making it impractical to do too frequently.”
Worse, it's not just a compute problem. Repeated fine-tuning doesn’t just overwrite old knowledge (catastrophic forgetting), it can actually shut down a model’s ability to learn from new data altogether.
3. What would it take to build AI that actually sharpens itself as it learns about you?
"As you work with a model day in and day out, the model becomes more tailored to your context, your use cases, your preferences, your environment. Imagine how much more compelling a personal AI agent would be if it reliably adapted to your particular needs and idiosyncrasies in real-time… it could create durable moats for the next generation of AI applications...This will make AI products sticky in a way that they have never been before."
Sounds great in theory. But how, exactly? No one really knows. Repeated fine-tuning isn’t just impractical—its repeated use degrades the model and can eventually turn it into total garbage. Maybe it’s time to admit: we need something new. Something fundamental is missing from today’s AI architecture.
Meanwhile, the next (might be multiple) release candidates are being developed/trained an tested for potential future production use.
e.g. When I did autonomous robotics, the sensor models had to be quite adaptive as less predictable environmental parameters such as lightning conditions, dirt, energy level and temperature could influence readings dramatically. These dynamic adaptations occur at runtime, sometimes by a fairly non trivial trained sensor model.
What you usually do not want is running an untested system that "freely" learns from presented data in a live production environment as that could lead e.g. to contextual over-fitting or destabilization and even subversion of the adaptive control processes.
Exceptions could be systems that have to operate in extremely dynamic and less understood environments, but where risks are bound and you can confidently implement guardrails to protect against excessive loss (e.g. HFT agents).
This statement reflects a common (and dangerous) assumption in today's AI culture—that one can foresee all possible future conditions at design time—knowing the unknown unknows. Zillow’s AI was also "declared fit"... until COVID flipped housing dynamics and cost them half a billion. Tiger Global’s $17B loss followed a similar trajectory—confidence in pre-deployment testing, blindsided by real-world shifts....I can go on. But the good news is some communities, especially those deploying AI in the real world, have started to recognize this. For example:
"Autonomous systems must be able to operate in complex, possibly a priori unknown environments that possess a large number of potential states that cannot all be pre-specified or be exhaustively examined or tested. Systems must be able to assimilate, respond to, and adapt to dynamic conditions that were not considered during their design... This 'scaling' problem... is highly nontrivial." — Institute for Defense Analyses (IDA)
Until the broader AI/ML culture internalizes this gap—between leaderboard AI (wins in pre-defined benchmarks) and real-world AI—we'll keep seeing deployed systems fail in costly, unpredictable ways.