GAK

[GAK @doublestands]

@bcherny The token spend explodes exponentially from 0 to 4 levels. Unless processes and human systems are optimized to absorb the value, it is not a good idea to spend $m at level 4. Ready to fire the majority of the team, only then go to 3 or 4

[GAK @doublestands]

@MichaelArnaldi Y-axis should be unemployability. Top devs cost a lot, so why will I use them at all in post-AI world.

[GAK @doublestands]

@mehulmpt thanks, all the best on your new location/life.

[GAK @doublestands]

@championswimmer First, if you are a good company with good values, you trust your employees that they are doing the right thing. Second, there have to be no incentives for tokenmaxxing. Given these conditions, if more tokens are burnt, it is a leading indicator of AI experimentation, no?

[GAK @doublestands]

@bindureddy Not sure if it will work, we had Bitnami OSS one-click deployments for a variety of these use cases, which never scaled. It is better to improve deep agents with more power so that they can create any custom tech stack the user wants.

[GAK @doublestands]

@fchollet Software is not dead, but margins will shrink from 50% to less than 5%. Software will just become like any other commodity, and there are business models that trade commodity but margins are lot less !

[GAK @doublestands]

@fchollet Most large SIs have stopped or significantly reduced hiring freshers from Indian campuses. As more budgets get allocated to ChatGPT enterprise licenses etc across large MNCs with dev shops, this trend will accelerate in 2024! Google "India developer jobs hiring freeze"

[Subbarao Kambhampati (కంభంపాటి సుబ్బారావు) @rao2z]

📢 ICYMI, here is the video of the keynote talk I gave last Friday at @SCAI_ASU's #AI day titled "Can LLMs Really Reason & Plan?". It is the most comprehensive of my talks on the topic to-date. Slides are available with the video.

https://t.co/9IkakLsaZo