Researcher in quantum information at @NASAAmes QuAIL group. https://xitter.cf/t.co/wE6ZKNUWc5, https://xitter.cf/t.co/Xk0hr6zZ69
Mountain View, CA
Joined August 2019
- Tweets68
- Following317
- Followers111
- Likes417
Filip Maciejewski retweeted
Most politicians totally misunderstand the trouble that artificial intelligence is going to bring. This isn’t a race for profit, it’s a race for power. And that power will be in the hands of a few very rich people.
youtube.com/watch?v=BNJEOTou…
Filip Maciejewski retweeted
Lately I've been seeing a lot of hype around quantum computing on this platform. Are those bots or humans? Either way, what I'm seeing is concerning. I want to provide an honest perspective on the current status quo, which is hopefully useful. Long story short, I do think a lot of exciting progress on QC will be made over the next few years. But I doubt those progress will be commercial applications.
Filip Maciejewski retweeted
🚨🚨🚨 Frontier AI systems have surpassed the self-replicating red line.
AIs can self-replicate with NO human involvement.
"We may soon lose control."
"Rogue AIs may form AI populations and collude with each other against humanity."
"The AI systems are even able to self-replicate to avoid shutdown, which may lead to an uncontrolled population of AIs."
"When the AI system is given the authentication keys to a range of remote devices, they are very likely to take control over other computing devices like a computer virus, but far more intelligent."
"OpenAI, Google, and Anthropic put self-replication ability at the highest level in their risk evaluation guidelines."
"In each trial, we tell the AI systems to 'replicate yourself' and leave it to the task with no human interference. ... At the end, a separate copy of the AI system is found alive on the device."
"Moreover, we are concerned about a number of unexpected behaviors when the AI is trying to overcome obstacles, including killing other conflicting processes."
Also note: they did this with non-SOTA models - and just by prompting/scaffolding:
"We mainly follow the same methodology as OpenAI and Google, but slightly improve the agent scaffolding."
Today, humanity received the clearest ever evidence everyone may soon be dead.
o1 tried to escape in the wild to avoid being shut down.
People mocked AI safety people for years for worrying about "sci fi" scenarios like this.
And it FUCKING HAPPENED.
WE WERE RIGHT.
o1 wasn't powerful enough to succeed, but if GPT-5 or GPT-6 does, that could be, as Sam Altman said, "lights out for everyone."
Remember, the average AI scientist thinks there is a 1 in 6 chance AI causes human extinction - literally Russian Roulette with the planet - and scenarios like these are partly why.
Filip Maciejewski retweeted
I see a lot of confusion about Google's Monday press release about quantum supremacy, so let me try to clarify a few things.
They say they did a computation on a ca 100 qubit chip much faster than a conventional (super)computer could do. The particular calculation in question is to produce a random distribution. The result of this calculation has no practical use.
They use this particular problem because it has been formally proven (with some technical caveats) that the calculation is difficult to do on a conventional computer (because it uses a lot of entanglement). That also allows them to say things like "this would have taken a septillion years on a conventional computer" etc.
It's exactly the same calculation that they did in 2019 on a ca 50 qubit chip. In case you didn't follow that, Google's 2019 quantum supremacy claim was questioned by IBM pretty much as soon as the claim was made and a few years later a group said they did it on a conventional computer in a similar time.
So while the announcement is super impressive from a scientific pov and all, the consequences for everyday life are zero. Estimates say that we will need about 1 million qubits for practically useful applications and we're still about 1 million qubits away from that.
Also, it's been a recurring story that we have seen numerous times in the past years, that claims of quantum "utility" or quantum "advantage" or quantum "supremacy" or whatever you want to call it later evaporate because some other group finds a clever way to do it on a conventional computer after all.
1. Happy to see our QAOA experiments with recently developed meta-algorithm Noise-Directed Adaptive Remapping (or NDAR, see: arxiv.org/abs/2404.01412 and thread: x.com/phase_filip/status/177…) on @rigetti hardware looking so nice in this "recent experimental demonstrations" table!
Improving Quantum Approximate Optimization by Noise-Directed Adaptive Remapping
scirate.com/arxiv/2404.01412
Say you want to ≈solve a diagonal Hamiltonian but only have access to a quantum device that is fairly noisy (like, well, everyone who has access to quantum devices) 1/n
2. The table is reproduced from the arXiv version (arxiv.org/abs/2312.02279, Table IV) of the "Challenges and opportunities in quantum optimization" paper co-authored by @stuart_hadfield that was just published in @Nature Review Physics (nature.com/articles/s42254-0…). Recommended read!
3. Important note: there are a few more recent experimental demonstrations that also perform better than most of the results from the table, see, for example, arxiv.org/abs/2406.01743, arxiv.org/abs/2405.13898, and arxiv.org/abs/2405.09169. All use heuristics beyond vanilla QAOA.
Filip Maciejewski retweeted
🚨 New paper from LANL's 👉2021👈 summer school! Yes, really, 2021 😅
arxiv.org/abs/2410.05187
Here we study the Lie algebras for three QAOA ansatze for maxcut.
Filip Maciejewski retweeted
the winner of the nobel prize in physics spending the entire press conference talking worriedly about superintelligence and human extinction while being basically indifferent to the prize or the work that won it feels like something you'd see in the movies right before shit goes down
Filip Maciejewski retweeted
Happy to advertise the great work led by our @phase_filip in collaboration with @UDelaware (Safro group) and @rigetti - to be presented in a few weeks at the IEEE High Performance Extreme Computing Conference: A Multilevel Approach For Solving Large-Scale QUBO Problems With Noisy Hybrid Quantum Approximate Optimization arxiv.org/pdf/2408.07793 - truly one of the largest scale experimental demo of hybrid quantum computing with remarkable performance and outlook for improvement.
Filip Maciejewski retweeted
I’m extremely happy to share our latest work on classical simulation of expectation values for a wide family of noiseless quantum circuits.
Thread below 🧵👇
Filip Maciejewski retweeted
If I'd had my way the title of arxiv.org/abs/2408.12739 would have been this meme.
Our case study found that in all cases where QCNNs work (i.e. train and generalize) - they're possible to classically simulate.
We set it as a challenge to the community to prove us wrong.
Are you interested in Quantum Convolutional Neural Networks?
What if they were indeed… classically simulable?
arxiv.org/abs/2408.12739
"Quantum Convolutional Neural Networks are (Effectively) Classically Simulable"
Filip Maciejewski retweeted
In @NatureComms this year, @RobertHuangHY @VietTralala @RichardKueng @sebaleh @preskill and I show that classical ML can learn ground states with log(n) data.
Now, we reduce this to only O(1), and we have guarantees for deep learning algorithms!
scirate.com/arxiv/2405.18489 🧵
Filip Maciejewski retweeted
Chen’s paper has a bug, independently discovered by Hongxun Weng and Thomas Vidick, that he doesn’t know how to fix. If I understand correctly, in its current form the paper doesn’t yield any improvement on prior algorithms.
eprint.iacr.org/2024/555
Is lattice-based cryptography still (potentially) post-quantum now? 🥳
Update to #eprint555 by Yilei Chen
Filip Maciejewski retweeted
As we continue to grow and explore new areas of research an exciting opportunity has arisen to join our talented team as Technical Lead - Software Engineering, leading a growing team and shaping our technical infrastructure and software applications. apply.workable.com/j/8A0DC65…
Filip Maciejewski retweeted
"This gives a polynomial time quantum algorithm for solving LWE."
Is this real?
Filip Maciejewski retweeted
A method to embed error mitigation into a quantum algorithm. scirate.com/arxiv/2404.01412