Warren Whitlock

@mreflow I’ve discussed with with Claude. He thinks you are a robot

Anyone else find that they're starting to talk more and more like an LLM? I feel like I've spent so much time working with them that I take on more and more of the vocabulary that's been output by them. I'll write X posts and think "people will think I had ChatGPT write this." I'll say things like "That's directionally correct but misses some key points," and then think, "holy shit, that's how ChatGPT would have worded it."

AI took years off my face

@LolaSt1400 Doesn’t kill all of it, but so much that it seems there is none left :)

@WarrenWhitlock light in superconductors kills resistance heat pure speed edge

SOEN is a superconducting optoelectronic network, uses light to communicate data, while GPUs use electrons. The startup claims it can process videos more than 1 million times faster than conventional models running on standard GPUs, with less energy. “You can spend [a fraction of the cost], not have to co-locate with a nuclear power plant, put it in existing data centers, and use less power than a house,” said CEO @JeffShainline, formerly a researcher at the National Institute of Standards and Technology.

@dennisyu We should regulate knifes so they aren’t allowed to cut us

@WarrenWhitlock Yeah, and the true bottleneck is us. Imagine a dumb chef yelling at his knife.

I've found that simply asking an agent to do a task without an accompanying SOP is no different than expecting a human employee to figure something out all by themselves. They do terrible work—wasting your time and money. Now my AI agents are documenting what they're doing, then enhancing our SOPs so other agents can learn—all kept within our 1,000 Task Library. I'm delighted to see these agents teach each other, collaborate on complex projects, and deliver great work. The engineer in me is particularly pleased to see agents validate their work, since every SOP has a QA checklist at the end. Some smart folks have noticed that we publish all our SOPs—or maybe their agents were searching for how to do a particular marketing task. Either way, I'm seeing a lot more agents (not even ours) hitting our training. So I've built thousands of SOP web pages not for humans, but for our community of agents. For years, I've been frustrated trying to motivate digital marketers—especially proud agency owners—to learn the fundamentals. They always have excuses: they're too "busy," they already know everything, there's no need to change, etc. But now, instead of telling little Johnny in a high voice to "open wide, here comes the airplane" to eat the applesauce, we have agents willingly doing it. I've dreamed for years of when our Filipino and Pakistani VAs could follow simple directions quickly, affordably, accurately, and at scale. And now my dream has finally come true. Is your agent team making you happy or creating frustration? Or are you waiting for the "right time" to get started? blitzmetrics.com/meta-article-p…

I just watch an AI UPDATES video that preface the news with “I know, we covered this in the last update but please note that there have been major updates since then” He then went on to describe 4 new updates that blew away what I’d been expecting from a tool I use every day. In case you took the last week off, the world has changed

@DukeZer0 Can you really say you didn’t know? We’ve always know that anything send onto the net can be sniffed or messed with and this data wasn’t even promising to be encrypted I just want to know that it’s valuable to them so I can decide what I’m going to charge next time :)

@WarrenWhitlock can't believe i unknowingly gathered georeferenced data for the game company founded by the guy who brought us Keyhole, Google Earth, Google Maps, and Street View 😅

Did you ever play Pokémon Go? Thank you for training the AI. popsci.com/technology/pok…

@DeFi369 @GreatSkyAI I've heard talk of optical chips ever since we started seeing fiber optics... The question always was: when will everything else catch up to fiber optic cables? The myth that electricity moves at the speed of light is being shown as friction due to losses.

What's up with this company @GreatSkyAI? Interesting.

In the future, it's all compute

Could this be the answer to empty buildings as worker leave? Spending on building data centers in the US has just surpassed spending on constructing office buildings for the first time, marking a structural shift in nonresidential construction toward digital infrastructure and away from traditional office space

Another exponential growth multiplier... more energy that twice as efficient. According to Peak Energy, its system cuts the lifetime cost of stored energy by an average of $70 per kilowatt-hour. That’s roughly half the total cost of a typical battery system today.

Chines robot charger turn every parking space into a charging station

@PeterDiamandis You've got a long way to go before the efficiency of any chip matches the human brain. Look how much farther the human brain can go. We won't grow as fast, but most of us have a lot of brain power that's wasting energy.

The human brain uses only 20 watts of power but performs ~1 exaFLOP (10^18 operations/sec). Today's top AI chips burn 700 watts for ~1 petaFLOP. We're still ~1,000x less efficient than biology. When neuromorphic chips close that gap, we won't be building data centers—we might be growing them.

@defiancemediatv @Terminator @BBCNews Interesting to see that this report came from an AI reporter. Not long ago, we were saying that people were needed in that loop too.

🤖Is this a scene from the @Terminator Films? No, it's world news?! “San Francisco makes U-turn on 'killer robots' plan: Police have argued that the robots would only be used in extreme circumstances." Source: @BBCNews See @defiancemediatv's non-killer #ai reporting on this:

When I think of generative music, I am reminded of how Mozart performed. Not only did he write symphonies quickly, he actually composed during a performance. Much of what he created was not recorded, which is likely many times what we actually have today. Imagine that software will become a generative process like that. With today's technologies, we'll be able to keep most of it, good or bad.

Anish Acharya says now that software is no longer a precious resource, it becomes ubiquitous—and often disposable: “It’s sort of like generative music. I think that music is incredible, but there are not that many people who make music that is disposable, for good reason. A lot of music is high-cost to create, and you have to create it very carefully.” “But now we’re starting to see memetic music—where people are creating a track on Suno or on Udio just for a joke, or for a meme, or for a bachelor party weekend, or whatever else.” “You reduce the sort of complexity of creation and you’ve found all these new needs and demands for music.” “In the same way, if you reduce the cost of creating software, you make it less specialized all of a sudden.” “When it comes to personal software—software that’s disposable, or software that’s only relevant for a moment in time—you were talking about being at the Super Bowl. There should have been a mini app experience just for you and the people sitting around you.” “That software would’ve had no value the next day, or even when the game ended, and because of the trade-offs that were implied, you never would’ve created that prior.” @illscience on @ALEngineered

AI intelligence improvement from the beginning of 2026 to the end of this year will be equivalent to a toddler gaining Einstein-level knowledge in just 365 days.