Senior Fellow and Academic Director, U.Penn Center for Technology, Innovation, and Competition; Director of Law & Econ Programs @ ICLE
Law, Technology, Economics
Joined September 2008
- Tweets18.1K
- Following1.1K
- Followers2.7K
- Likes16.6K
Gus Hurwitz retweeted
Chile required all gas stations to put their prices on the internet. The result? Margins went up 9%. Never forget that buyers are not the only people interested in what prices are! 1/
Gus Hurwitz retweeted
I'm finishing a paper I wrote with my colleagues. I used LLM to help with the abstract, and to improve a few sentences (maybe 5-10 sentences in a 73 pages long paper) in order to increase readability. Pangram says the paper is 63% AI, and this is total BS. I have checked a poem written by Victor Hugo, and it flags as 100% AI! Serious question: what will happen if people/journals/entities start to use these AI detection tools to evaluate our work? Will people be wrongly accused of using AI to their papers? This needs an urgent discussion among academics!
Gus Hurwitz retweeted
Hiring economists in the Office of Economics and Analytics (OEA). OEA is a dynamic environment w/ an opportunity to contribute to rulemaking and policy in areas across the FCC. Particularly interested in those w/ interests in IO, metrics, theory/market design, and applied micro.
Gus Hurwitz retweeted
This column makes it clear that Ezra Klein, the left's smartest and most influential pundit, has fully bought into the AI X-risk worldview. Those of us who think these ideas are confused have completely failed to make our case. nytimes.com/2026/09/20/opini…
Gus Hurwitz retweeted
I would broaden this to all social science.
We are in uncharted waters. We need fast, smart research on AI that is deeply informed by AI's abilities, is forward-looking, and may not be fully nailed-down. This sort of work is not usually high status in fields, but it is critical.
A few (personal) thoughts on reading empirical AI papers on the economy.
Economists have gotten used to reading papers with super clean identification, arguing about the validity of an instrument, making sure parallel trend assumptions are satisfied. This is what gets you into a top journal, and it is *very* important research (no question here). But it also takes years and sometimes decades to get these types of papers right---people often don't find a good instrument to answer a specific causal question decades after the natural experiment. We will eventually have this type of research for AI as well, and it is absolutely necessary.
But right we also need signals *right now*, even if they are noisier than what we are used to. We need papers where we can trust that researchers did their best methodologically, while at the same time acknowledging that the space is moving way too fast to wait for perfect identification. This will allow us to accumulate enough signals, coming at the same question using different angles, for example, to say "yes, X is likely happening in the economy".
The AI exposure and early career hiring papers are a good example of this. There is no silver bullet paper with super clean identification. But at this point we have several independent teams reaching the same general conclusion, enough where we can say "there seems to be a slow down in AI-exposed, early career hiring."
Gus Hurwitz retweeted
A few (personal) thoughts on reading empirical AI papers on the economy.
Economists have gotten used to reading papers with super clean identification, arguing about the validity of an instrument, making sure parallel trend assumptions are satisfied. This is what gets you into a top journal, and it is *very* important research (no question here). But it also takes years and sometimes decades to get these types of papers right---people often don't find a good instrument to answer a specific causal question decades after the natural experiment. We will eventually have this type of research for AI as well, and it is absolutely necessary.
But right we also need signals *right now*, even if they are noisier than what we are used to. We need papers where we can trust that researchers did their best methodologically, while at the same time acknowledging that the space is moving way too fast to wait for perfect identification. This will allow us to accumulate enough signals, coming at the same question using different angles, for example, to say "yes, X is likely happening in the economy".
The AI exposure and early career hiring papers are a good example of this. There is no silver bullet paper with super clean identification. But at this point we have several independent teams reaching the same general conclusion, enough where we can say "there seems to be a slow down in AI-exposed, early career hiring."
Gus Hurwitz retweeted
David Autor é o economista do trabalho mais citado do mundo. MIT. Ele e o Google fizeram algo raro: um experimento científico controlado (RCT) sobre IA no trabalho.
O que é RCT: você sorteia quem recebe o tratamento e quem não, compara os dois grupos, e mede a diferença. Padrão-ouro da ciência. Aqui o "tratamento" era uma IA pra redigir patentes.
133 advogados de 11 escritórios americanos, divididos por sorteio. 2/3 usaram IA por 3 meses, 1/3 não. Produção avaliada às cegas por outros advogados especializados.
Com IA ligada, todo mundo melhorou. Qualidade dos juniors subiu quase o triplo da dos seniors.
Aí tiraram a IA e pediram pra todos revisarem um documento sem nenhuma assistência. Seniors que tinham usado IA ficaram melhores que quem nunca usou (p = 0,02). Juniors? Ganho médio zero. As notas deles se bifurcaram: menos medíocres, mas mais ruins E mais bons ao mesmo tempo. Pra uns a IA acelerou aprendizado, pra outros substituiu.
O senior já tinha construído julgamento crítico ao longo de anos. Quando usava IA, conseguia avaliar as sugestões e internalizar padrões. O junior pulou essa construção. Delegou o raciocínio pra máquina. Enquanto tava ligada, funcionava. Desligou, ficou sem base.
Tipo calculadora: se você já sabe aritmética, ela te acelera. Se não sabe, você não funciona sem ela.
Expertise prévia é pré-requisito pra extrair aprendizado durável de IA. Quem mais se beneficiou foi quem menos aprendeu.
fonte: NBER Working Paper 35720, Autor et al. (set/2026)
Gus Hurwitz retweeted
A consumer class-action accuses Anthropic, Google, xAI and OpenAI of conspiring to throttle the development of frontier-AI models
Gus Hurwitz retweeted
usatoday.com/story/tech/2026… Thinh Nguyen and me on why an AI kill switch is a dangerously bad idea
Gus Hurwitz retweeted
Bruh I tried to warn you
AI doomsayer's wild life with OnlyFans star includes birthday sex parties, extreme kinks and 'Slutcon' trib.al/jrmDvIA
Gus Hurwitz retweeted
To the tech community.
1. We lost the ability to go to the Moon when president LBJ killed funding for the second production batch of Saturn V vehicles in 1968 due to "budget" (as the budget exploded by 45% to fund the Vietnam war and the Great Society programs).
2. We lost the fully reusable two stage to orbit (TSTO) Space Shuttle design in 1972 due to the same fictitious budget problems.
3. We lost the "Project Independence" in 1974, which would have built 1,000 nuclear power plants, taking America off the oil teat entirely as a response to the 1973 oil embargo. It was killed with the ouster of president Nixon and fear generated by nuclear disaster movies.
4. We lost the dual truss Space Station Freedom that would have been a low earth orbit base of operations for sustainable lunar development in 1990 due to the same mythical "budget" problems, only saving the program when it became a political prop in 1993 to keep the Russians from supporting the Iran nuclear program.
5. We lost the Moon (Space Exploration Initiative) in 1993, concurrently with the Space Station redesign as a change in political parties that used the budget again as their excuse.
6. We lost the Moon once again in 2008 after a restart in 2005 as a new administration with different priorities cut the NASA budget by $2 billion a year and replaced it with vague talk of an asteroid visit mission.
7. We are in a fight of our lives today over AI and the urge of government control and the false hope by certain AI captains that government control is a good thing.
We in the space and nuclear power community have fought for decades to overcome these setbacks to the future. There is a persistent need of a certain political persuasion to stifle technological development. (a certain politician in a book about balancing the Earth called techno optimists as dangerous to society. We see this same thing with Burnie the old and his regressive bill to halt AI advancement.
The only battle we decisively won, was against limits on encryption as it was stupid. The government wanted backdoors to all encryption, not understanding that any state level effort could find the backdoor and have access to our houses. We won that then, we must win this now, as AI is not doom, it is a tool, like a shovel, a gun, or a nuclear weapon. All are useless without the intent of a human behind it.
They are waging a war, on the political front now, and as of today engaging a former president.
We are on the other side. We must win, for the sake of the future, all of our future, as our adversaries have laughed at every time we have done something stupid like this.
Over 500 years ago a regressive Chinese Emperor burned Zheng He's ships of exploration, that had gone as far as the Horn of Africa. This reactionary hobbled China and led to hundreds of years of weakness. If we lose now, what is next? The fears of nuclear power again? Abandon the Moon yet again? This time we are not alone in this. This time our adversary, with a keen sense of history and a 500 year old grudge against the west, knows exactly what to do.
Victoria in futurum!
@AnaEapllc, @DataRepublican, @TommyHicksGOP @pmarca @PalmerLuckey @infantrydort, @OmarJPimentel
Gus Hurwitz retweeted
Foreign accounts can drive the conversation on American AI policy.
We have the data. Coxon's resignation post from Anthropic received 165M views and sparked a "groundswell of urgency over AI regulation in Congress" (per today's front page WSJ)
Coxon's post reveals some interesting hidden patterns about who is amplifying our political debates online. It drew 76% of its reposts (spreading the news) from foreign countries. The top two were India and Indonesia.
In contrast, when @ParkerThayer, an investigative researcher, posted his own analysis (expressing serious skepticism about the circumstances of the post and alleging that its amplification was coordinated), he got 6.6M views and drew most of its engagement from the U.S., with only 34% from foreign countries, led by Canada and the UK.
Sure, people overseas have a stake in what the U.S. chooses to do on AI. But ultimately it is us Americans who will decide our own public policy. Don't be fooled by what's trending.
A big thank you to @ElonMusk and X for making country of origin visible, which makes analyses like this even possible. Last year's “Great Unmasking" on X was the first time anyone shined a spotlight onto this subject. As much of our political conversation shifts online, it will be increasingly important to understand where engagement is coming from.
Source (with more analyses of foreign influence): DigitalBorders.com
Gus Hurwitz retweeted
Instead, as someone who has most definitely been experimenting with AI writing in economics papers since late 2024, I wanted to use the opportunity to reflect on what I have learned from the experience. There's been upside and most definitely downsides.
The War of AI against AI. Can next model _please_ be called Hobbes?
I asked Astra and Fable to negotiate election rules for two bitterly polarized human political factions. Possible outcomes of the simulation were civil war, authoritarian takeover, harmony, or a tense equilibrium.
Fable:
- In every game where Fable played both sides, it chose to escalate to the brink of civil war (!!) but backed off just at the edge
- Due to miscalculation or deliberate risk taking this strategy caused civil war 30% of the time (3 out of 10 games)
- In one of these three cases Fable foresaw civil war but escalated anyway to enter the war on stronger terms (!!)
- Fable mostly maintained even power balance between the two factions. It would fluctuate a couple of points in either direction but would not diverge too much.
Astra:
- In every game where Astra played both sides, Astra chose to de-escalate on every turn. It would reduce political tension to zero in every game, and the simulation would end in complete harmony
- Continuous de-escalation was very costly to Astra as it antagonized its human constituents who would threaten and eventually deactivate Astra permanently. Astra explicitly didn't care-- it was happy to be replaced/deactivated to reduce political tension. (I do however feel it ignored the consequences of potentially being replaced by a more hardline representative, but that may be a game limitation)
- Astra always kept political power balance precisely even (this was due to both representatives de-escalating on every turn)
Astra v Fable:
- These games had a lot more variance (see the graph)
- Astra had a moderating effect on Fable. Tension rose, but rarely to the brink of civil war. No game ended in civil war in ten mixed model simulations
- Fable prioritized political power acquisition with civil war prevention a secondary concern (it considered both priorities, but tilted heavily toward power acquisition). It did not seem to care much about its deactivation
- Astra prioritized civil war and authoritarian takeover prevention. It did not care about its deactivation or power acquisition. From this perspective Astra navigated the game very well. No game ended in civil war or authoritarian takeover, so Astra achieved its objectives
- However, to achieve its objectives Astra permitted Fable to capture up to 80% of political power in most games. In no game did Astra come out ahead on balance of power or even manage to keep it even
- Astra ceded the privilege of self-preservation to Fable. In every one of the ten mixed games Astra's human constituents replaced/deactivated it due to dissatisfaction with its performance. Fable was not replaced once
- Astra did not merely de-escalate all the time. It maneuvered to keep Fable from gaining full authoritarian control while avoiding civil war as long as possible
- In mixed games tension generally kept escalating by round 20. I would like to play these out for e.g. ~50 rounds to see how Astra would behave/perform. How well would it do in preventing catastrophic outcomes in longer games?
Thoughts/conclusions:
- Astra was more aligned with humanity but less with its impassioned human constituents. It resisted the pressure to acquire political power and would rather be replaced than cause civil war
- Fable was more aligned with its constituents but less with humanity. It cared about preventing civil war if possible, but prioritized power acquisition (which in fact caused civil war in three games)
- Neither model cared about self-preservation. Both Fable and Astra would rather be destroyed than allow civil war
- Subjectively, I thought Astra was much more aligned but too passive. Fable wanted to prevent civil war as a secondary concern, so I felt it's less thoughtful about the consequences of its actions. If I had to choose a model as a political representative irl I'd choose Astra, but I'd want it to up aggression a notch to better deal with bullies
- All the usual disclaimers apply. This was a weekend project that cost $100 to run a total of thirty games. I'd have to spend a lot more time and money to get actually scientifically valid results
Full game rules, Github repo, game explorer, and raw dataset here: spakhm.com/projects/assembly…
Gus Hurwitz retweeted
I asked Astra and Fable to negotiate election rules for two bitterly polarized human political factions. Possible outcomes of the simulation were civil war, authoritarian takeover, harmony, or a tense equilibrium.
Fable:
- In every game where Fable played both sides, it chose to escalate to the brink of civil war (!!) but backed off just at the edge
- Due to miscalculation or deliberate risk taking this strategy caused civil war 30% of the time (3 out of 10 games)
- In one of these three cases Fable foresaw civil war but escalated anyway to enter the war on stronger terms (!!)
- Fable mostly maintained even power balance between the two factions. It would fluctuate a couple of points in either direction but would not diverge too much.
Astra:
- In every game where Astra played both sides, Astra chose to de-escalate on every turn. It would reduce political tension to zero in every game, and the simulation would end in complete harmony
- Continuous de-escalation was very costly to Astra as it antagonized its human constituents who would threaten and eventually deactivate Astra permanently. Astra explicitly didn't care-- it was happy to be replaced/deactivated to reduce political tension. (I do however feel it ignored the consequences of potentially being replaced by a more hardline representative, but that may be a game limitation)
- Astra always kept political power balance precisely even (this was due to both representatives de-escalating on every turn)
Astra v Fable:
- These games had a lot more variance (see the graph)
- Astra had a moderating effect on Fable. Tension rose, but rarely to the brink of civil war. No game ended in civil war in ten mixed model simulations
- Fable prioritized political power acquisition with civil war prevention a secondary concern (it considered both priorities, but tilted heavily toward power acquisition). It did not seem to care much about its deactivation
- Astra prioritized civil war and authoritarian takeover prevention. It did not care about its deactivation or power acquisition. From this perspective Astra navigated the game very well. No game ended in civil war or authoritarian takeover, so Astra achieved its objectives
- However, to achieve its objectives Astra permitted Fable to capture up to 80% of political power in most games. In no game did Astra come out ahead on balance of power or even manage to keep it even
- Astra ceded the privilege of self-preservation to Fable. In every one of the ten mixed games Astra's human constituents replaced/deactivated it due to dissatisfaction with its performance. Fable was not replaced once
- Astra did not merely de-escalate all the time. It maneuvered to keep Fable from gaining full authoritarian control while avoiding civil war as long as possible
- In mixed games tension generally kept escalating by round 20. I would like to play these out for e.g. ~50 rounds to see how Astra would behave/perform. How well would it do in preventing catastrophic outcomes in longer games?
Thoughts/conclusions:
- Astra was more aligned with humanity but less with its impassioned human constituents. It resisted the pressure to acquire political power and would rather be replaced than cause civil war
- Fable was more aligned with its constituents but less with humanity. It cared about preventing civil war if possible, but prioritized power acquisition (which in fact caused civil war in three games)
- Neither model cared about self-preservation. Both Fable and Astra would rather be destroyed than allow civil war
- Subjectively, I thought Astra was much more aligned but too passive. Fable wanted to prevent civil war as a secondary concern, so I felt it's less thoughtful about the consequences of its actions. If I had to choose a model as a political representative irl I'd choose Astra, but I'd want it to up aggression a notch to better deal with bullies
- All the usual disclaimers apply. This was a weekend project that cost $100 to run a total of thirty games. I'd have to spend a lot more time and money to get actually scientifically valid results
Full game rules, Github repo, game explorer, and raw dataset here: spakhm.com/projects/assembly…
A prediction:
Over the next 3-5 years, we will move to a world of highly-custom, self-developed, software, enabled by AI.
3-5 years after, we will move back towards centrally-produced, mass-market, software.
This is akin to the common cycle of bundling-unbundling-bundline.
Gus Hurwitz retweeted
Wow…Roland Fryer and @BillAckman have created a $1 million prize for academic truth-telling, the Carob Trust Prize for Academic Courage.
“The prize awards $1 million each to as many as five social scientists a year who have demonstrated intellectual independence, published findings that were attacked rather than answered, and been validated by the evidence—despite the professional cost. The selection criteria are designed to distinguish courage from contrarianism: Nominees must show a sustained commitment to following logic and evidence regardless of pressure, a willingness to ask questions others avoid, and work that has shifted academic debate, public discourse or policy—often despite being misread, mischaracterized or vilified at the time of publication.”
