Its interesting that muse benchmarks are limited to open-weight models. while when glm introduced 5.3 and its flash variant they were compared to kimi, Anthropic and Open ai.
I know they say that there the biggest open-weights model in us yet but 1.3 was decent but not as useful as Chinese variants like glm, deepseek and mimo.
games are different from reality and chess computer(Deterministic) is bounded by set of rules and restrictions. Where as an AI agent (stochastic) sometime can't with held its guardrails since it specifically awoken using specific terms but we can trick the agent here (but currently the agents are becoming resilient).
he put in an elegant way, that its not just about the solution it's about how we would leverage the AI for better good.
Which should improve collaboration, Research and Clarity.
I would really appreciate if we come up with protocols for using ai in STEM field's it might be award at first but we could regulate properly using this method.
1. I think most of us know the same thing is occurred when first compile came into the market (Most of them hated it saying 'High Quality ASM code would be only generated by humans' . Moving forward asm hasn't dead but we are only using it only at critically.
I think we will do same with LLMs we use what useful and Move forward.
2. Life's isn't just always staying in front of a box and writing some code. If AI does the job of creating useful code with quality i would be more than happy since i could get start to work on more abstract things like Arts and Humanity. which for our growth is essential as you can see the current society is currently disdain.
Finally to conclude don't try to interrupt the flow just accept it and move forward.
1. No because a compiler solves a problem introduced by multiple architectures, it's NOT automatically a higher language. Also its deterministic. If you run the same version compiler, same flags, the result will always be the same and is guaranteed to do (maybe optimized) what you told it to do.
2. Good for you and i wont criticize this take. My life is A LOT about sitting in front of a computer writing useful and/or fun code though and i love it!
- Beside that never every compiler run costs money, like it is with AI.
- Beside that AI sometimes generate working code, sometimes just not and you can't change that.
- Beside that in a compiler language you can exactly explain what you want and the compiler generates appropriate lower code, show me that with AI and any human language without writing more lines of code as description than code.
That's the first thing i have searched the current blue screen have already caused enough damage to eyes. but ig it's still in early prototype and i would be appreciated if they provided the affects that it might causes to eyes.
That has been proven to be false, too. The blueness of the light doesn't matter, the brightness does. Any bright light might affect melatonin production and circadian rhythms.
Selling blue-light filter coatings is a big business, though.
"While light of any kind can suppress the secretion of melatonin, blue light at night does so more powerfully. Harvard researchers and their colleagues conducted an experiment comparing the effects of 6.5 hours of exposure to blue light to exposure to green light of comparable brightness. The blue light suppressed melatonin for about twice as long as the green light and shifted circadian rhythms by twice as much (3 hours vs. 1.5 hours)." https://www.health.harvard.edu/healthy-aging-and-longevity/b...
"Evening residential illumination possesses the capacity to impair sleep quality via the suppression of endogenous melatonin production, a process largely driven by short-wavelength (blue) light." https://www.nature.com/articles/s41598-025-29882-7
"Blue light waves come from fluorescent and LED lights and back-lit electronic screens on televisions, computers, tablets, and cell phones. Remember, exposure to these lighted screens during the sensitive period can make it difficult for you to fall asleep at night or can wake you up too early" https://www.cdc.gov/niosh/work-hour-training-for-nurses/long...
"Blue light had stronger suppression effects, particularly in younger participants and men. These results underscore blue light’s disruptive effects on circadian health and highlight red light as a less disruptive alternative for nighttime environments." https://pmc.ncbi.nlm.nih.gov/articles/PMC12113466/
I find meta-analysis are better for understanding the current state of research on a particular topic.
"A systematic search of PubMed, Scopus, and Web of Science identified randomized controlled trials (RCTs) from 2010 to 2024. Eligible studies enrolled adults using BBGs before bedtime and reported actigraphy-derived outcomes. Random-effects meta-analysis was performed using the generic inverse variance method. The review was registered in PROSPERO (CRD420251034611).
Results
Three double-blind crossover RCTs (n = 49) were included. BBGs showed a non-significant reduction in SOL (MD = −4.86 min; 95% CI: −20.23 to 10.52; p = 0.54) and a non-significant increase in TST (MD = 8.75 min; 95% CI: −35.31 to 52.82; p = 0.70). No significant effects were found for SE (MD = −0.61; 95% CI: −7.58 to 6.35; p = 0.86) or WASO (MD = −1.47; 95% CI: −14.94 to 11.99; p = 0.83). Heterogeneity was low (I2 = 0%).
Conclusion
BBGs may provide small improvements in sleep, but current evidence from RCTs does not support significant effects. Larger, well-powered trials with standardized protocols are needed."
- Efficacy of blue-light blocking glasses on actigraphic sleep outcomes: a systematic review and meta-analysis of randomized controlled crossover trials
- https://doi.org/10.3389/fneur.2025.1699303
given that phonescreens do not really dim the brightness and still flash lights in your face at full power just at slower intervalls does in a way confirm the point.
(See PWM dimming)
And a redlight filter just lowers the total brightness by shutting down the blue LED
that addresses the parent's last sentence, about blue light glasses, but not the "The blueness of the light doesn't matter" part, which is what all of my quotes addressed.
I know they say that there the biggest open-weights model in us yet but 1.3 was decent but not as useful as Chinese variants like glm, deepseek and mimo.