@MattWalshBlog I send content that I choose for my children to their tablet lock screens and they have to interact with it to gain access, and that's only when I allow them to use it, which isn't always
Being a parent of a teenager in the modern digital environment means begging your kid to play Minecraft or do *something* even slightly less brain-rotting than watching short video content from Spotify, which is what they're started doing because you've blocked all YouTube content from the XBox devices at the router level and turned off the WiFi and limited the carrier data throughput after 9PM for their smart phones, which they legitimately need because that is how their coaches communicate with them about cheer routines or band practice
And if you dare to complain about how annoying it is to have this second job as an IT admin just to keep your kids from guzzling down the worlds shittiest content, half the internet calls you a bad parent for not locking them down more and the other half calls you a bad parent for being a content Nazi
If manufacturers really want to impress me with smart home appliances, give me a documented local API.
I don't need another mobile app, another cloud account, or another subscription just to control something sitting three metres away ffs
It's all about finding the tasteful touch points that make sense. Most of our actual homeschooling is analog, however, we have very beautiful, logical, and efficient uses of AI where it makes sense.
@Arkypatriot I'd allow AI to assist me in homeschooling, but not take it over completely. It can help outline lesson plans and pretty up assignments to help me keep him interested, but the entire point of homeschooling is to take an active role in educating your child.
I should add, I don't have this problem much with anything code related because I'm using @poteto's pstack verification idea, so, this seems to happen more with non-programming tasks. I don't think we should have to write an independent verification skill for everything, e.g. checking ride stats from an exercise app, so maybe a general claim verification skill can work? It feels like this should be solved at the harness level though.
I've been having this problem too. It doesn't look like there's a "pre-response hook" we can use and bot memory fails to solve this. I just had my bot create an "inspect before claiming" skill. It shouldn't be necessary and I don't know if it will help, but it's worth a try.
These little things keep happening consistently over the last few days. I really can’t believe how disappointing Grok Bot has gotten so quickly. I think Muse might actually be more intelligent now. I know these are first-world problems, but it’s all relative, and this product has noticeably gotten worse recently. As someone who loves this product dearly, I can’t help ranting about it because I want them to fix it.
I completed a bike ride and nudged my bot to check the stats and give me a review. It always used to just know to check whenever I said anything that implied I finished a workout. Now it’s braindead apparently and needs me to spell it out.
This morning, I weighed myself on my smart scale and it said I was 10lbs lighter than the day before. So obviously the scale was wrong. My bot was like “I’m going to record this datapoint and log it as suspect.” This is even after I noted that the scale was obviously busted. I had to tell it explicitly to omit the datapoint.
These are small things but they totally ruin the experience and erode your trust in the models. And this stuff was not happening at all before. Never.
This plus usage nerf is just so wild. The magic is gone
AI can easily boost the productivity of an individual by immeasurably high amounts, but it can't boost the productivity of an organization one iota without leadership.
No, mine is a way poorer version of what you're talking about. I just open Claude, Grok, or Codex in directory and tell it what to do. Mostly it's just just tracking business expenses by indexing receipts, with the knowledge that we'll do a schedule C come tax time. I haven't felt the need to promote it to fulltime bot yet. I will if my need grows, but if it comes to that, I'll probably do it as a @NousResearch Hermes.