By returning detailed docs for exactly what the AI is coding at the time, it greatly reduces the the likelihood it will make a mistake. It moves from a recall from the training data problem, to a transcription problem.
This looks promising. I'm looking forward to exploring this further and seeing its practical applications in streamlining document-related workflows. It's always refreshing to see new solutions emerge that potentially shift how we approach common challenges.
Why would China try these postures when the opposite country is neck deep in handling a pandemic and one which originated from its land. Is this any time for a bloody battle or power trips?. May be there is more to it than the reasoning you mentioned. Internal frictions are starting to get visible against premier due to the way he handled the covid and how much repair it has caused China globally. May be he is trying to do all this antics to shore up nationalism and create a new problem/diversion which makes covid related ones to take a back seat.
> India doesn't stand a chance in an actual conflict
If its a full blown war then may be yes. But two nuclear powers going all out and all alone will not happen so others will step in. It would complicate things for China. Ideally it should be conventional and a very short one. The rules for engagement can't be escalated, that means China has to work with severe geographic disadvantages.
Chinese investment money is everywhere so they try to control the narrative. I have seen for sometime that on HN, we are open to criticize but don't take a standpoint unless its something to do with western government.
I think this is due to the flame war or rating system of HN, where active discussions are relegated to oblivion. Instead of trusting biased and funded media - we here need to introspect, without us being silenced.
May be we are mostly left so we don't have a much stronger opinion unless its right - is that the case?. Just curious - why the slack?.
If yours is a small team and if many things are keeping your team busy - never address a technical debt head on until you can afford the luxury or until the business demands warrant for it. Most engineering team focus too much on technical elegance and go overboard. Rewrite culture can become nasty if unchecked.
> But there is a trick you could use if you're not a programmer: visit a top computer science department and see what they use in research projects
Please don't shoot yourself with this suggestion. From my personal experience Academia is so distant from what is relevant or practical to the industry.
Somewhat related, most academia is not focused on source code quality or its suitability for posterity. If you absolutely need to mime or plunder academia to succeed in your startup, pair that with a software industry veteran who can scale and secure your technology operation.
I have the opposite opinion. US funding sources for science and engineering tend to be very government-heavy rather than industry-based; but in my experience those government grants, particularly DoD, emphasize research with specific practical ends. The obvious exception is NSF, but Europe also has similar obvious exceptions at about the same rate.
The big difference between the US and Europe, however, is that universities in the US depend on grants to fund PhD students. Many (most?) European countries have national funding programs for their PhDs, which frees up both the students and their advisors to work on whatever fancies them regardless of practicality. France is particularly notorious for this. US academics do not have this luxury.
In my field, by far the lion's share of impractical (Fun! I love it! But impractical) work comes from Europe.
Yeah this is specific to Europe, in particular Germany, where companies such as BMW would send engineers to work on a defined research project at a university, and then adopt the research if it pans out. In the US there’s far more leeway, and part of being a good PI is stealthily inserting yourself own agenda into the DOT grant proposal, and those in the DOT will give you the leeway knowing that’s “how it works”
I guess this varies a lot. To my shame I don't have hard numbers (! I really should have.)
Personally I've worked on a half dozen global research network projects bringing together both public and private funding and 10-15 organisations, mixed universities, research institutes, and small to behemoth companies.
The target is always some wide ranging new radical technology platform and they spawn a bunch of startups and spin off projects.
Some targets can turn out to be too difficult, or simply wrong, but I've never seen a project that hasn't lead to at least one startup / product / new technology.
Eg: Start out looking at nanoparticle functionalisation, then stumble on a very useful sensor application or targeted molecular delivery.
Eg: Work on autonomous air sensor, then find simple off shoot application for tuberculosis using a completely different technology.
Eg: Build a medical laser for transdermal drug delivery then find applications in dental work and scar tissue removal.
Eg: Design a flexible heterogeneous compute cluster for hpc application, then get customers who want it for the low administration cost.
You start by bringing some very competent and diverse people together to tackle some tricky and new problem. Then after a while you'll have found a bunch of peripheral questions and applications for one thing or other you create in the project.
People talking and sharing experience and knowledge will almost always find new stuff out in the dark. If you're working on the edge of knowledge there is almost always interesting stuff hiding just outside the light of your personal flashlight. Bringing people together in multidisciplinary projects is amazing.
I don't see any tech solutions mentioned out here. At least i would like to take a shot.
Not relying on government is the way to go.
I think we should build a social network for community welfare which can prepare us for something like this in future. Where people build communities among themselves. Key point being internal resource availability to be exposed to the close knit network and an external pool exposed to an radius. Keeping everything to the close knit shouldn't be encouraged and there should be a proportion. Ideally the network should be split when its crossing a proportion.
How do we keep up with something like covid-19 in future?. I don’t think relying on government is the way to go.
Why not build a social network which helps to build close knit communities and it can have a cascade effect on the whole populace as a side effect("good"). Some of the key points for it to work being:
# Pool resources before hand so that the whole system is not suffocated.
# Internal resources exposed to close knit.
# External resources exposed to a radius.
# Cap on internal vs external resources.
# Split the community into smaller ones when it reaches a proportion.
# Point system to motivate communities.
# Ability to mobilize resource across communities.
This need not be the case with just a pandemic but can help us tide over other eventualities as well.
you need more connections than what a single AWS CLI process open to saturate network on a big box. You can achieve the same by doing some "xargs | aws cli" trickery but then error handling becomes harder.
Our S3 client just handles multiple worker processes correctly with error handling.
Its hard to estimate without breaking it down to smaller tasks and this process is laborious. This process assumes we are tackling stuff in which we already have expertise.
The rule of thumb i use - anything which takes more than 3 days needs to be broken down further into tasks. If any task takes more than 3 days during execution then it needs to be broken down to more tasks.
