Aš studijuoju dirbtinį intelektą ne tam, kad sukurčiau sekančią self-driving car. Ne tam, kad AI pakeistų žmones darbo rinkoje. Mano tikslai didesni. Viliuosi, kad pavyks pasiekti kada nors tokį advanced AI lygį, kuris nebepabraukinės mano vardo ir pavardės worde ar rašant email
@eglepug Kantrybės. Mums ant pavadžio išmokti vaikščioti reikėjo apie dviejų-trijų metų. Net ir dabar ne idealiai eina, bet ženkliai geresnė situacija.
@levelsio@octal Regarding ACs -oat new buildings have them. The only ones that are missing are those where the builder aimed to save costs. However, people install them just fine. You should visit Baltic states + Poland if you want to see how 21st century Europe looks like
@levelsio@octal That's because you look at western Europe countries that are still stuck in 20th century. In Lithuania for every house built pass 2016 it is mandatory to have a ventilation system. Since 2018 it should be an active air control system. It is enforced by law and actively checked
@yongfook@jessethanley Not necessarily. If your highest churn rate on low tier customers, spending more on new high-tier user acquisition would make more sense. It all depends acquisition costs vs your churn rate
@levelsio If back in 2006 when the internet was really slow 100ms delay lead 1% conversion rate decrease for Amazon, then the impact in 2025 should be much higher. Ref: https://t.co/ZDHIAUyQrA
@kuchin@jobergum@aiDotEngineer@saranormous You build evals and then hope for the best. That's why all evals utilize OTel for request tracking and for classical ML it was not needed on a such granular level. You inspect all your historical agent steps and then iterate to improve