Mark @Markdotmd

Applied AI United States Joined April 2008

Tweets

2K
Followers

806
Following

919
Likes

7K

Mark @Markdotmd

6 days ago

@andrewute1 Utah replaced 5 OLs. It doesn’t have the power run game success it’s had in the past. I think they are adjusting to the current OL capabilities and using lots of misdirection is probably required to balance other weaknesses

0 0 0 40 0

View Details

Mark @Markdotmd

6 days ago

@GoodSirHamlet It’s a roster issue. It’s not scheme or effort or coaching. Utah doesn’t have guys talented enough on the D line to sack the QB. Or at least not developed yet. Guessing it gets solved next season. Until then it’s still a 9-10 win team but not yet championship ready

1 0 2 35 0

View Details

Mark @Markdotmd

6 days ago

Utes have two major problems and neither is fixable in season: power run game (50% conversion ) and pass rush (52nd). These are roster problems, not stuff to fix in practice. The D looks great for 2 quarters each week then then backfield gets tired and we get gashed. On offense when they need to re-take momentum and control clock they struggle a bit. This is still a 9-win roster. A very good team. But winning a conference championship is the wrong expectation of this roster IMO

1 0 2 119 0

View Details

Mark @Markdotmd

a week ago

Maybe we always feel this way. In the 90's it was the birth of connectivity and in the 2020s it's AI. The 2030s likely will be robotics. But it's always technology. From the book The 5th Discipline.

0 0 2 39 0

View Details

Mark @Markdotmd

2 weeks ago

@DCinvestor I assume walled gardens emerge and only those with proof of humanity can participate. Some will also narrow it down to a white list of humans who can join. Will take a few years but we will adjust.

0 0 0 47 0

View Details

Mark @Markdotmd

2 weeks ago

@tszzl Why don't we just follow the pattern we used to solve human alignment?

1 0 0 53 0

View Details

Mark @Markdotmd

2 weeks ago

Talking to LLMs all day trains you to easily see any kind of original thinking at all on this app pretty fast.

0 0 1 32 0

View Details

Mark @Markdotmd

3 weeks ago

Cost per successful AI turn seems like the best way to measure ROI on our tools right now. Token spend is a meme.

0 0 1 37 0

View Details

Mark @Markdotmd

3 weeks ago

People chose email products largely based on broader ecosystems (Google-Gmail, Microsoft-outlook etc). Email is email. I think intelligence converges on the same thing. Your preferred model will be the one that ties an ecosystem together.

0 0 1 46 0

View Details

Mark @Markdotmd

3 weeks ago

Somehow the people using math to build more sophisticated LLMs are confused that LLMs are just really complex linear algebra equations. The internet provided inputs large enough that LLMs are finding more and more novel connections. Never confuse math and data surface areas with self awareness. LLMs are incapable of reasoning without input tokens. They don’t scheme in between turns. If the inputs stop, so do the outputs. A conscious entity wouldn’t operate that way.

0 0 0 23 0

View Details

Mark @Markdotmd

3 weeks ago

@citrini Hahahaha shorting the IPO

0 0 0 97 0

View Details

Mark @Markdotmd

4 weeks ago

@NickNemo17 Any chance you have any invites available for it?

0 0 0 62 0

View Details

Mark @Markdotmd

a month ago

@deobfuscations I had early access but for some reason I cannot deposit without it getting rejected. I cannot withdraw unless I use a debit card which has fees. Customer service is worthless and will not fix it. I have the card. But until I can deposit and withdraw without issue it's pointless.

1 1 4 270 0

View Details

Mark @Markdotmd

a month ago

testing @PrismML new bonsai model and adding @typesafeai Jev agent in front of it to see if we can make it even smarter

0 0 1 44 0

View Details

Mark @Markdotmd

2 months ago

@icobeast Tomb forks also emerging

0 0 1 157 0

View Details

Mark @Markdotmd

3 months ago

@markkaplan20 Thank you. I’ll look into the other scan for better insight. Plus the other markers

0 0 0 9 0

View Details

Mark @Markdotmd

4 months ago

@BorisMPower If no tokens are being fed in, the model isn’t thinking, dreaming, or planning its own goals. No tokens in nothing out. How can consciousness be applied to those conditions?