Mark Histed

"There is ample evidence that people form strong, lifelong preferences during their late teens and early twenties. (What’s your all-time favorite album or movie?)"

Lots of interesting stuff in here. Other parts bear on the future of regulation and agencies too. We have to make things work better.

I just watched this around 10 times

Leading with the crown, looks like

“We’re going to continue to hire, but we’re also going to continue to build agents,” he said.

As usual, the article is more nuanced than the headline. But in this case I fear the headline is going to mislead readers. The story mostly talks about how McKinsey is using AI alongside workers.

Sounds like a banned word list!

“Another popular agent checks the logic of a consultant’s arguments, verifying the flow of reasoning makes sense.”

No! Bad!
This isn’t going to work, not in general.

“McKinsey is rapidly deploying AI agents… to assist consultants in building PowerPoint decks, taking notes and summing up interviews and research docs for clients. The most-used bot helps employees write in a classic “McKinsey voice”—language the firm describes as sharp, concise and clear”

Great!

lol CHIRP

generative AI making errors is Very Bad for McKinsey.

A friend who is now a senior partner at a consulting firm told me years ago “You (associates esp) can never make a single mistake in a client meeting, because they’re paying for your stellar reputation.”

(Not directing that at you, @jeffjarvis.bsky.social. Thanks for posting it.)

The industry it misunderstands is mgmt consulting. The firms that can pay McKinsey’s high fees are looking for smart people, sometimes for insight— but often for reputational heft and down-low info exchange.

I hate to bang on about the same thing, but:
AI is Not coming for the consultants, @wsj.com.

Generative AI can do amazing things. Coding! Translation! Mediocre text generation at scale!

But none of these are why clients pay McKinsey the big bucks. This fundamentally misunderstands the industry.

“democracy, the worst possible system except for all others that have been tried”

We have a long history to draw on here

And yes, the substack/subscription model is not sustainable. Because it produces only a few big winners, but it's too much work for even a superstar to do real reporting on it for the long term. You need a media outlet for that, at least a small org.

This is why we need public funding for journalism, and we should find ways to include podcasters and influencers that are reaching audiences with good information.
We can structure this public funding to favor small institutions like @404media.co, @51st.news. They'll all be doing short video soon

👇

I'd say the purpose of the sherman and clayton acts was to prevent these very kinds of attacks on rule of law, and in many ways they worked! We just got complacent and let the laws be rewritten.

But I'm just a 'normie liberal' who's watching the rules-based system inside gov't implode

Agree lack of enforcement helped get us here—you're right to point to Sherman and Clayton Acts, which have been gutted the same way as the VRA.

But I disagree w/Ganz it's inherent to the system: it's rather an attack on the system from the outside.

thanks!

Yes. I’m really interested in how AI is being used by professional mathematicians.

Solving the IMO problems is one thing. Doing new math is different.

@teorth.bsky.social has organized some conferences on this and written about it. I should look into it more.

For ex

bsky.app/profile/teor...

Yes. I’m really interested in how AI is being used by professional mathematicians.

Solving the IMO problems is one thing. Doing new math is different.

@teorth.bsky.social has organized some conferences on this and written about it. I should look into it more.

Try Deep Think in the Gemini app Deep Think utilizes extended, parallel thinking and novel reinforcement learning techniques for significantly improved problem-solving.

New release from Google

“Because it can reason through highly complex problems, Deep Think can be a powerful tool for researchers. It can help formulate and explore mathematical conjectures or reason through complex scientific literature, potentially accelerating the path to discovery.”

“I will never let the potential for personal recognition, profit or advancement cause me to act in a way that violates the public trust in science or in me as a scientist;”

The same impulses that make people identify with tugboats and snowplows or any toy with a face kicks in with a machine that can introduce itself and simulate a conversation with you. Are there problems with the conversation? Yeah, but we're built for papering over those problems without noticing.

AI to Assist Mathematical Reasoning: A Workshop The National Academies of Science, Engineering, and Mathematics are hosting a virtual workshop on the topic of “AI to Assist Mathematical Reasoning” from June 12-14. The tentative progr…

There's some debate in math whether and how AI will replace or assist mathematicians in consequential ways.

But I think it's pretty clear that most of political science is not that; neither is journalism; that new reasoning and confirmable ideas are not things that gen AI does now

Working with AI: Measuring the Occupational Implications of Generative AI Given the rapid adoption of generative AI and its potential to impact a wide range of tasks, understanding the effects of AI on the economy is one of society's most important questions. In this work, ...

Here's the paper

arxiv.org/abs/2507.07935

(Sigh.)
#ai #neuroai

Working with AI: Measuring the Occupational Implications of Generative AI∗ Kiran Tomlinson1 , Sonia Jaffe1 , Will Wang1 , Scott Counts2 , and Siddharth Suri1 1Microsoft Research 2Microsoft Abstract Given the rapid adoption of generative AI and its potential to impact a wide range of tasks, understanding the effects of AI on the economy is one of society’s most important questions. In this work, we take a step toward that goal by analyzing the work activities people do with AI, how successfully and broadly those activities are done, and combine that with data on what occupations do those activities. We analyze a dataset of 200k anonymized and privacy-scrubbed conversations between users and Microsoft Bing Copilot, a publicly available generative AI system. We find the most common work activities people seek AI assistance for involve gathering information and writing, while the most common activities that AI itself is performing are providing information and assistance, writing, teaching, and advising. Combining these activity classifications with measurements of task success and scope of impact, we compute an AI applicability score for each occupation. We find the highest AI applicability scores for knowledge work occupation groups such as computer and mathematical, and office and administrative support, as well as occupations such as sales whose work activities involve providing and communicating information. Additionally, we characterize the types of work activities performed most successfully, how wage and education correlate with AI applicability, and how real-world usage compares to predictions of occupational AI impact.

Table 3: Top 40 occupations with highest AI applicability score. Job Title (Abbrv.) Coverage Cmpltn. Scope Score Employment Interpreters and Translators 0.98 0.88 0.57 0.49 51,560 Historians 0.91 0.85 0.56 0.48 3,040 Passenger Attendants 0.80 0.88 0.62 0.47 20,190 Sales Representatives of Services 0.84 0.90 0.57 0.46 1,142,020 Writers and Authors 0.85 0.84 0.60 0.45 49,450 Customer Service Representatives 0.72 0.90 0.59 0.44 2,858,710 CNC Tool Programmers 0.90 0.87 0.53 0.44 28,030 Telephone Operators 0.80 0.86 0.57 0.42 4,600 Ticket Agents and Travel Clerks 0.71 0.90 0.56 0.41 119,270 Broadcast Announcers and Radio DJs 0.74 0.84 0.60 0.41 25,070 Brokerage Clerks 0.74 0.89 0.57 0.41 48,060 Farm and Home Management Educators 0.77 0.91 0.55 0.41 8,110 Telemarketers 0.66 0.89 0.60 0.40 81,580 Concierges 0.70 0.88 0.56 0.40 41,020 Political Scientists 0.77 0.87 0.53 0.39 5,580 News Analysts, Reporters, Journalists 0.81 0.81 0.56 0.39 45,020 Mathematicians 0.91 0.74 0.54 0.39 2,220 Technical Writers 0.83 0.82 0.54 0.38 47,970 Proofreaders and Copy Markers 0.91 0.86 0.49 0.38 5,490 Hosts and Hostesses 0.60 0.90 0.57 0.37 425,020 Editors 0.78 0.82 0.54 0.37 95,700 Business Teachers, Postsecondary 0.70 0.90 0.52 0.37 82,980 Public Relations Specialists 0.63 0.90 0.60 0.36 275,550 Demonstrators and Product Promoters 0.64 0.88 0.53 0.36 50,790 Advertising Sales Agents 0.66 0.90 0.53 0.36 108,100 New Accounts Clerks 0.72 0.87 0.51 0.36 41,180 Statistical Assistants 0.85 0.84 0.49 0.36 7,200 Counter and Rental Clerks 0.62 0.90 0.52 0.36 390,300 Data Scientists 0.77 0.86

Microsoft just released a study w/ the 40 jobs most and least likely to be replaced by AI.

On the most likely: Mathematician, political scientist, and journalist.

Sorry to the authors.
But anyone who thinks these fields will be supplanted (helped) by AI knows little abt both these fields, and AI.

you're lucky to have such a great candidate for mayor!