I don’t know enough about this to raise it with Axim, but aren’t the models listed in the policy among the most expensive to run? Presumably those costs would fall on us as the providers, right?
Yes, though you’re free to use them as much or as little as you like. Even not at all.
Makes sense, but isn’t it still quite limiting if you want to use AI assistants without spending a lot? It’s a bit like saying, “Sure, you can travel, but only by helicopter.”
Costs introduce limitations, yes. There are ways you can use AI assistants here which don’t require spending money beyond current hardware if you wanted to run a local model, for instance. They just can’t include code contributions. Orchestration workflows are still on the table, summarization and question-asking is still on the table, etc.
Each team member gets to pick their tools and decide what it’s worth to them. That’s been the case with our workstations, our IDEs, our network connections, etc. Whether or not to use AI coding assistants to write code is also a choice, and which provider to select is one as well.
Note that we all wrote and contributed code perfectly fine before these tools. No one here is green enough in their practice where they couldn’t, so they can also be judicious about token use when they choose to use it.
Just avoid getting Nestle’d if the business model changes, losing your ability to write code while the costs become out of reach. You should be able to continue working even if your favorite AI service is down.
3 Likes
I am going to use that analogy now XD thank you for pointing it out 
1 Like