Google is not comparing the costs or usability of chatbots versus traditional search. They are comparing someone else getting chatbots to replace a significant amount of their search traffic compared to Google doing the same and cannibalizing their own traditional search. It's not a gross margin question for Google but an existential question.

AI that can build itself would be a major development in the history of technology -- one that could bring enormous good for the world in science, healthcare, and beyond

Indeed!

If it were possible to effectively slow the development of this technology to give ourselves more time to deal with its immense implications, we think that would likely be a good thing

No, it would not be. I want more people's lives — my own included — improved by those developments. And I want it yesterday.

Imagine Wright brothers sabotaging airplane-development, because it would allow people to travel too far too fast? Or the early automakers fretting over "implications" of using internal combustion engines for personal vehicles — because millions of grooms and coachmen would lose their jobs?.. Electric lamp? Wow, nice — but what about the candle-makers?

AI can easily write the lean for any proofs they do. The problem is that lean is missing large parts of established mathematics.
My current lean project has to cite all of these externally due to missing lean support.

    Functional equations / means
    - lit_aczel_1948 — Aczél: symmetric + homogeneous means power/quasi-arithmetic mean (the CES forcing)
    - lit_aczel_1966_weighted — weighted Aczél characterization (weighted means)

    Fixed-point / topology
    - lit_brouwer_1911 — Brouwer fixed point
    - lit_cellina_approximate_selection_1969 — approximate selection (closed-graph correspondences); with Brouwer
      Kakutani
    - lit_glicksberg_1952 — Glicksberg fixed point (infinite/Bayesian games)
    - lit_berge_maximum_theorem_1959 — Berge maximum theorem (upper-hemicontinuity)

    Probability / large deviations
    - lit_sanov_1957 — Sanov / method of types (large-deviation rate = KL)
    - lit_fisher_tippett_gnedenko_1928 — extreme-value theorem (GEV limit laws)
    - lit_kolmogorov_1931_fokker_planck — Fokker–Planck diffusion equation

    Optimal transport / matching
    - lit_sinkhorn_1967 — Sinkhorn matrix scaling (entropic OT)
    - lit_lp_strong_duality_1951 — LP / transportation strong duality (Gale–Kuhn–Tucker)
    - lit_entropic_penalty_cominetti_sanmartin_1994 — entropic-penalty -convergence (T0)
    - lit_gale_shapley_1962 — deferred acceptance produces a stable matching
    - lit_gale_shapley_proposer_optimal_1962 — proposer-optimality of deferred acceptance

    Stochastic calculus / PDE
    - lit_ito_1944 — Itô's lemma
    - lit_black_scholes_pde_solution_1973 — closed-form solution of the Black–Scholes PDE
    - lit_liouville_dirichlet — Liouville/Dirichlet (harmonic-function / PDE result)

    Dynamical systems
    - lit_saddlenode_passage_time — saddle-node "bottleneck" passage time / (Strogatz/Fenichel)

In a way, Amazon "improved" upon one-click and invented zero-click.

It's not psychosis at all. The CEOs likely know more than the average slashdotter. What might look like psychosis is much more reasonably explained by the chase for stock bonuses. AI-heavy companies like Nvidia and OpenAI would lose most of their value if AI popped, and their CEOs would lose many billions of dollars. They need to promote a rosy AI future because their they would drop from insanely rich to merely extremely rich otherwise. Even companies like Google, Apple, Meta, Amazon, and Microsoft feel the need to promote AI because they envy the stock returns of the pure-AI companies.

What do the CEOs know? That some AI domains are already extremely successful (e.g., factory monitoring and maintenance), selectively successful (e.g., financial analysis, ads), or soon to be successful and life-altering (e.g., medical diagnosis, medical treatments). Some domains are mostly not currently successful and have murky outlooks (e.g., software coding). Some domains are intended to enable future domains (e.g., frontier models) or pursued as halo projects. The people who hail AI as the precursor to the singularity and the people who deny all current AI successes are both blind.

Exactly. This is a very good sign. Despite overt and implicit opposition from the current US government, wind and solar continue to grow. That suggests strong inherent economic motivation.

"Siri should finally be able to understand more personal context, have on screen awareness, and be able to take action in apps for you. This'll finally be made possible thanks to Apple's new partnership with Google, where Apple will be using Gemini-diffused models hosted on Private Cloud Compute to power Siri"

So, basically Apple is marketing existing tools as Apple and Siri? The good thing about selling AI is that most consumers don't understand what's happening underneath the chatbot interface, so Apple can claim to be making AI innovations, and some people will believe it.

If the entry bar is a string of literals representing keywords to search for, then the response should be the old list of links to associated pages. However, if the entry bar is a conversational request, then treating "disregard" as a request rather than a literal search term is reasonable. Of course, it might be argued that no one would actually bother to approach Google to tell them to disregard a request. However, that's perhaps a matter of conversational protocol. What would a human do if someone approached and said, "Disregard"? There are likely other ambiguous or awkward instances of conversational protocol that would be misinterpreted.

The main problem is that the entry bar is intended to field both search requests and conversational requests at the same time, but there is currently no syntax or protocol to distinguish the two. Simply relying on context is insufficient because humans would have the same parsing challenges. Maybe putting quotes around literal search terms is one way, but quotes already have special meaning in search terms.

LPR surveillance is unconstitutional.

No, it is not. There is no such article in the Constitution.

If they want to use LPR information, then make it a warranting process.

Ah, you're implying, the 4th Amendment covers license plates? No, it doesn't — the license is outside in plain sight. If I can legally see it, I can record it.

Now, the very requirement to have the license plate in the first place — that seems quite bogus to me. Not unconstitutional — just wrong. There is no argument for license plates on personal vehicles on the road, that wouldn't also apply to actual persons on the same road...

We can blame social media, billionaires, idiots, etc. However, one big thing that needs to be called out more is an increasingly authoritarian US government that is on an audacious and so-far unchecked binge of canceling. This Friday, Stephen Colbert is being canceled because Trump wanted it.

AI already works, but only in some areas like factory management, financials, medical, and ads. Those areas have already seen implementation, added functionality, and profits. However, Meta doesn't sell into those areas, aside from ads. Meta seems to be flinging people around trying to find something that sticks. However, Meta's AI incompetence doesn't erase the already existing AI successes in other companies.

Is this true of MS programs that don't require writing a thesis or research or publications? Many MS degrees, even from name universities, are purely taking more courses.