DEVOURED
The rise and fall of Jev hype
The 'Jev' hype cycle highlights an industry pattern of using jargon-heavy, manipulative marketing to obscure the lack of fundamental innovation in new AI models.
What: Alex Ewerlöf dissects the launch of 'Jev', a decision model that was hyped as a 'System One' AI without hallucinations, but proved to be a limited, API-locked classifier lacking actual reasoning capabilities.
Why it matters: The industry's cycle of 'Peak of Inflated Expectations' followed by rapid disillusionment is a symptom of a market starved for genuine breakthroughs, forcing vendors to rely on celebrity name-dropping and manufactured drama.
Deep dive
- AI marketing patterns: name-dropping (e.g., 'former ChatGPT creator'), jargon usage, and artificial scarcity.
- 'Jev' is a non-autoregressive decision model, not a general-purpose reasoner.
- 'No hallucination' claim in decision models often just means strict schema compliance, not logical truth.
- The true 'bottleneck' for AI is 20-Watt efficiency, not parameter size.
- We remain far from 'System Two' reasoning, continuous real-time learning, and real-world embodiment.
Decoder
- System One/System Two: Based on Daniel Kahneman's work; System One is fast/intuitive/automatic, while System Two is slow/logical/deliberate.
- Pareto curve: The set of models that provide the best possible performance for a given compute budget.
Original article
Full article content is not available for inline reading.
Explore all 69 articles in today's digest
Fresh dev news, AI-digested daily. Never miss a trend at devoured.fyi