News on China's scientific and technological development.

latenlazy

Brigadier
I had a friend who worked directly under Andrew Yang in Baidu’s AI lab (this was back in 2016) and his conclusion then was that Baidu’s NLP and DL systems, including for applications such as self driving cars, were much more advanced than what they saw from Google, at least back then. I think it’s very easy for people to make too big a deal out of what sometimes amount to AI gimmicks. While these things make for a good show they don’t always end up being developed into meaningful applications. From what I’ve observed ChatGPT seems impressive from a subjective human experience standpoint because of its uncanny stylistic mimicry, but I’m not so sure that has much meaning for applications that involve the advance of physical systems. The amount of progress being made here is probably smaller in substance than it feels in conversation.
Sorry that should be Andrew Ng* oops.
 

measuredingabens

Junior Member
Registered Member
I just asked this ChatGPT this question:

How much citric acid to add to water to get the pH of distilled vinegar?

The reply:


But no idea if the answer is accurate as I know nothing about chemistry.
I spent way too much time checking ChatGPT's anwers and equations, so here goes. The values for the dissociation constant, H+ concentration and citric acid molar mass are correct, as are most of the explanations. That said, the equations are really off. My main contentions would be steps 3 and 4. I plugged in the numbers and the end result very much didn't match the bot's. The result I got from (1 ÷ (1.58 × 10^-3)) ÷ 7.4 × 10^-4 is 0.00855285665 or 8.55 x 10^-3. I don't know where ChatGPT pulled 150mol/L or 0.67% from but it doesn't match the result to the equation.

For step 4 it should be 1.58 x 10^-3 / 3 rather than 3 x 1.58 x 10^3 when following the bot's logic, as citric acid's triprotic structure would mean that its concentration should be a third rather than three times that of the H+ concentration as otherwise you will end up with around nine times the desired H+ concentration. I also don't know where it got 0.85g of citric acid from, since when I followed the results given by GPT's equations and converted the moles of citric acid per litre (3.16 x 10^-5) to grams ((3.16 x 10^-5) x 192.12) it was a few orders of magnitude off the answer ChatGPT has given (0.00607g as opposed to 0.85g).

In short, the explanations are mostly sound, but the equations and numbers amount to bullshit.
 

broadsword

Brigadier
I spent way too much time checking ChatGPT's anwers and equations, so here goes. The values for the dissociation constant, H+ concentration and citric acid molar mass are correct, as are most of the explanations. That said, the equations are really off. My main contentions would be steps 3 and 4. I plugged in the numbers and the end result very much didn't match the bot's. The result I got from (1 ÷ (1.58 × 10^-3)) ÷ 7.4 × 10^-4 is 0.00855285665 or 8.55 x 10^-3. I don't know where ChatGPT pulled 150mol/L or 0.67% from but it doesn't match the result to the equation.

For step 4 it should be 1.58 x 10^-3 / 3 rather than 3 x 1.58 x 10^3 when following the bot's logic, as citric acid's triprotic structure would mean that its concentration should be a third rather than three times that of the H+ concentration as otherwise you will end up with around nine times the desired H+ concentration. I also don't know where it got 0.85g of citric acid from, since when I followed the results given by GPT's equations and converted the moles of citric acid per litre (3.16 x 10^-5) to grams ((3.16 x 10^-5) x 192.12) it was a few orders of magnitude off the answer ChatGPT has given (0.00607g as opposed to 0.85g).

In short, the explanations are mostly sound, but the equations and numbers amount to bullshit.

Ok, thanks. But do you know the answer? Need not be absolute precision, but it will save me the expense of buying a pH meter.
 
D

Deleted member 23272

Guest
Yes, that’s a reasonable answer. The US simply has more wealth than China,they can afford to test all kind of new stuff even without clear commercial value at the beginning.
China is at a different developmental and wealth accumulation stage than the US
Both of y'all could've made it simpler, ie. China is still a middle income country. I mean think about it, despite being the second biggest economy in the world, China's HDI level and per capita GDP is is still on the level of countries like Thailand, Malaysia, Turkey, Mexico, and Azerbaijan. Yet just because of its sheer size China has many regions now with similar wealth and living standard levels comparable to the West, and thus money to burn on an educated work force to create an economy mirroring that of the developed world. Like, in actuality a few of the aforementioned countries actually do exceed China in HDI or per capita GDP levels, or both. Yet, do any of them bost being included in the top 20 of the global innovation index, or having companies like Huawei, DJI, CATL, TikTok, or a leader in any emerging technology like Quantum Communication?

China punches above its weight, but let's not have illusions about where it stands in the world. It has limitations, but those limitations also mean it has room to grow. Eventhough a certain country nowadays and a few of its allies would love to stop that from happening.
 

measuredingabens

Junior Member
Registered Member
Ok, thanks. But do you know the answer? Need not be absolute precision, but it will save me the expense of buying a pH meter.
Honestly, it's been a while since I did pH and pKA calculations so I'm rusty in that area. Unless you have someone qualified to check the results of my calcs (or just do them for you) I really don't think you should be using them. That said, you'll need to buy a pH meter, I'm afraid. Unless you have the disposable income to buy everything needed to control temp, water quality (using deionised water for the solution is an option depending on what you are using the solution for, and you'll need it to rinse the meter anyway), citric acid purity (some batches are better or worse than others) you're going to need a pH meter and do things by trial and error. There are simply too many variables at play that you'll need that pH meter. In all the chemistry and biochemistry labs I did at uni there was a lot of adjustment needed beyond the initial calculations for experiments needing a specific pH. Also, buy a weak base, because there's a decent chance you'll overshoot the amount of citric acid needed and have to adjust the pH back up.
 
Last edited:

tphuang

Lieutenant General
Staff member
Super Moderator
VIP Professional
Registered Member
Please, Log in or Register to view URLs content!
Sinopec with China's first methanol H2 station in Dalian
can produce 1t of 99.999% high-purity Hydrogen fuel/day
Considering the cost of production, storage and transportation, it can reduce the cost of hydrogen by more than 20% compared with traditional hydrogen refueling stations.
At present, Sinopec has built nine hydrogen fuel cell hydrogen supply centers across the country, and launched a 10,000-ton-level photovoltaic green hydrogen demonstration project - Sinopec Xinjiang Kuqa Green Hydrogen Demonstration Project Construction, the company's first self-developed megawatt The first-grade proton exchange membrane (PEM) electrolysis water hydrogen production device has also been successfully put into operation in Yanshan Petrochemical.

so basically, moving around hydrogen is hard, so they instead move liquid methanol and then produce hydrogen fuel from that. Seems kind of wasteful in energy, but I guess this is the most economic solution.

Please, Log in or Register to view URLs content!
Today, about 60% of the world’s methanol is manufactured and used within China—making the country the global leader. It’s currently used mostly in plastics manufacturing.

Traditionally, methanol is produced from fossil fuels like coal and natural gas, but it can also be made from renewable resources like agricultural waste.

Please, Log in or Register to view URLs content!
They are also testing out Carbon capture to methanol plant.

None of these things are really commercially viable right now, so all the methanol are probably produced with coal and natural gas at the moment.
 

BlackWindMnt

Captain
Registered Member
My take is that instead of copying off stackoverflow, most programmers will use ChatGPT for, ahem, reference.
Yeah I expect some vs code or nvim popup where you can insert your ChatGPT query and it will show your result in a new window, tab or copy it to your clipboard. "How transform one object into another object using linq in c#" then it will spit out some references.
Its faster and less likely to distraction you given you don't have to use a browser to go to google or stackoverflow.

And then get stuck in debugging hell when they can’t trace back the genesis of the logic it spits out.
Yeah Chatgpt will probably not help you setting up, testing or debugging a distributed system.
I also wouldn't be surprised if ChatGPT would shut itself down if it saw some of the old enterprise code i have seen, my soul has been scarred by some of it:p
 

xypher

Senior Member
Registered Member
I spent way too much time checking ChatGPT's anwers and equations, so here goes. The values for the dissociation constant, H+ concentration and citric acid molar mass are correct, as are most of the explanations. That said, the equations are really off. My main contentions would be steps 3 and 4. I plugged in the numbers and the end result very much didn't match the bot's. The result I got from (1 ÷ (1.58 × 10^-3)) ÷ 7.4 × 10^-4 is 0.00855285665 or 8.55 x 10^-3. I don't know where ChatGPT pulled 150mol/L or 0.67% from but it doesn't match the result to the equation.

For step 4 it should be 1.58 x 10^-3 / 3 rather than 3 x 1.58 x 10^3 when following the bot's logic, as citric acid's triprotic structure would mean that its concentration should be a third rather than three times that of the H+ concentration as otherwise you will end up with around nine times the desired H+ concentration. I also don't know where it got 0.85g of citric acid from, since when I followed the results given by GPT's equations and converted the moles of citric acid per litre (3.16 x 10^-5) to grams ((3.16 x 10^-5) x 192.12) it was a few orders of magnitude off the answer ChatGPT has given (0.00607g as opposed to 0.85g).

In short, the explanations are mostly sound, but the equations and numbers amount to bullshit.
It is actually a common issue with ChatGPT - it produces a lot of bullshit that it is told confidently and looks reasonable for a layman. This is the reason why e.g. ChatGPT-generated posts were even banned (or are still banned) from StackOverflow - they can easily confuse a novice with a total bullshit answer. Same goes for many other tasks, riddles, etc. - you often need to spend quite some time carefully crafting your response and correct the mistakes before ChatGPT outputs something good. It is still can be useful but it is not "AI taking over jobs" level as clueless media and laymen are trying to present.

There are some tasks where ChatGPT can shine really brightly - e.g. creative writing. The model can "hallucinate" really well under vague constraints and can produce really spectacular results - basically, it can write very believably and eloquently about 'nothing' or abstract things, just like most modern 'journalists' and other bullshitters which is why this section of people is freaking out right now. At the same time, this exact same property of "hallucinating" leads to the non-sense equations, incorrect code, made-up bibliographical lists, etc. when you need factual or strongly constrained information.
 
Top