Hacker Newsnew | past | comments | ask | show | jobs | submit | sabj's commentslogin

If you research the ways data can be leaked out of an LLM interaction you can see some more subtle cases.

What if I ask it to replace every vowel in the secret code with an emoji from a library? Or translate it into binary? Etc.

Whether or not this implementation is narrow (by design), there's a good reason to invest in this kind of safety and security space.


You're right, that is the hard part of LLMs and why LLMs aren't catching on broadly as a UI alternative beyond tech demos.

Probably the only true alternative is to limit user input to something structured and verified.

Until LLMs improve, their use in sensitive applications don't make sense and this product does little to improve that.


Points not covered in this piece, but worth noting: (1) If a counterparty is interested in you, it can help accelerate valuable partnerships you care about (2) You may want someone to invest in you, not to buy you; or you may want them to invest in you in the future

Now, for CorpDev, a conversation about investment is always a sliding scale... n% (invest) <--> 100% (buy you). But whether it's from the corporate balance sheet or as a referral to the corporate VC arm, there can be value there, and value in the relationship building, depending on circumstances.

As with so much, the risk is not knowing what you want and getting carried along by the process -- lettings things "happen" to you. If you have a conversation with corpdev, you're trading some information and receiving some information. Is that ride worth the price of admission? You have to decide based on the circumstances. This piece has an edge that helps to provoke and draw attention to the themes (don't let others shape the narrative of your business engagement) and that's fine.

There are places and indicators that a CorpDev conversation is 200% M&A, and there are times when it has more BizDev dimensions. Many companies want to develop partnerships first to determine if a potential acquisition is accretive. And, those partnerships can actually be valuable to small companies, even if there have tricky strings that can trip you up. ¯\_(ツ)_/¯


It took a long time. I was there.


One reason for these tough requirements is that a very large portion of U.S. passenger rail routes go over track owned (and used) by freight rail companies. Therefore, passenger train collisions and other such risks are weighed not only against other passenger trains, but also heavier freight trains.

This is one reason why U.S. trains have (relatively) 'harsher' requirements than other countries where track is not shared to such a high extent.

Separately, this degree of track-sharing is one of the reasons why average passenger rail journey speed has decreased in many places over a 100-year time period!

Obviously, in both cases, there are many many factors involved - but, an important element.


Have enjoyed watching Adioso develop and evolve over time, continuing to improve. Good job team. But, for me, the lack of additional airline data renders it completely unusable still, to have so many missing - I just can't get the whole picture price OR schedule-wise.

I always felt that travel search was a big pain point that I was invested in solving. It also has some very obnoxious data lockdowns after scale. Anyone here have a ready blueprint for some great resources to test and hack around different travel search and booking engines with helpful APIs? I know some offer odds and ends...

Everything is running extremely sluggishly for me right now - including the wingtip time reported on the page? - dunno if HN is a contributor but FYI! Totally unusable even vs. doing month-wide ITA Matrix searches.


It seems as if they only search Expedia, judging by the graph's label at the top. Even still, it's missing many cheap and budget airlines.

Whilst I commend them for letting me search over a period of time using natural language, I'd rather spend the time searching individual days on a search engine that has more sources.


Not just Expedia at all.

The progress indicator above the calendar chart shows that we hit Expedia, Cheapoair and Adioso's own Wingtip engine.

Wingtip has most low-cost airlines across most of the world. See http://adioso.com/help/coverage


+1 for Ryanair (even though they are an awful company). You are also missing norwegian which pretty much is the go-to budget airline for anyone living in the Nordics/Scandanvaia (& Baltics).


No RyanAir is a bit of a problem, and I'm failing to get EasyJet results to show on some routes (e.g. LGW-TLL)


Cheap ones from Mexico: Vivaerobus, Volaris and Interjet are missing too.


A fun article whose real value will likely be missed by many here who will be rushing, misguided, to defend their ideas of 'technology' or 'progress' from attack.

The points that Taleb is making here are best discussed, not by considering technology as artifact alone, but technology for what it really is - a set of artifacts, systems, processes, and artifices that take place in a social milieu and as a two-way relationship with society and civilization. To think otherwise is naive folly, from my perspective. (Please see: http://en.wikipedia.org/wiki/Social_construction_of_technolo... for a primer on this, it's fun reading if you're unfamiliar).

Technophiles (myself included!) will often project things out because they fail to understand that technology does NOT just show up and leave an imprint on the world, deus ex machina style. Instead the world shapes technology back. You can see this for any kind of technology, from the bicycle (Bijker's famous example, shown in the wikipedia article) to the Internet, which has its present design because of the real fears / threats / motivations that abounded during its establishment.

It's the (mostly) impossible nature of this dis-entanglement that throws people off track. In some world of scientist-kings of technocratic technophile dictatorships, yes, we'd have moon bases by now. But that's not how it works, and as a result people - messy, annoying humanity - get in the way. For better or, for often, worse.

The future WILL be cool - I'm still an optimist. But it will be cool in different ways, and maybe for different reasons, than many here on HN might imagine.

Since this thread is already a bit long let me instead just close with a great quote that I reference often. Relevant here.

The plain message physical science has for the world at large is this, that were our political and social and moral devices only as well contrived to their ends as a linotype machine, an antiseptic operating plant, or an electric tram-car, there need now at the present moment be no appreciate toil in the world, and only the smallest fraction of the pain, the fear, and the anxiety that now makes human life so doubtful in its value. There is more than enough for everyone alive. Science stands, a too competent servant, behind her wrangling underbred masters, holding out resources, devices, and remedies they are too stupid to use. – H.G. Wells, A Modern Utopia (1904)


To quote from the Wikipedia article:

"Advocates of SCOT—that is, social constructivists -- argue that technology does not determine human action, but that rather, human action shapes technology."

Sounds like another case of people not understanding feedback loops - that A can both determine, and be determined by B.


Obviously you're going to see a gradient, but I don't think most would contend human action isn't shaped at all by technologies in ways that are often unexpected. Plus, you get a sort of chicken-and-egg problem at some levels...

But people very, very commonly fail to see the embedded social factors in technologies, and are very quick to ascribe autonomy to technology, which is in my mind fallacious. You see this when the news has stories like, "Technology just keeps advancing. How will it change our lives next?" instead of understanding it as part of a relationship between people, their environment, and their artifacts, technology gets put on some magical self-propelling trajectory that just isn't true. Yes, Moore's Law is great - but it's a social/human driver, as is Intel's Tick-Tock, not something innate in technology!


The market problems faced by a phone and a tablet can be pretty distinct. If Microsoft is making a Windows Phone, I hope it serves to drive innovation and platform success, instead of just scaring away OEMs and cannibalizing what nascent platform there is.

If it's a foolhardy or hubristic attempt to grab imagined profits without consideration for carrier and other channel and customer obstacles, well... I'd regret the untimely / early demise of an innovative platform.


As lionhearted said, a mix of good points and bitterness, but 甜 and 酸 always go well together, don't they?

I think the most poignant section was this:

>A deal had been struck. Deng had promised the Chinese people material wealth they hadn’t known for centuries on the condition that they never again asked for political change. The Party said: “Trust us and everything will be all right.” >Twenty years later, everything is not all right.

But I would disagree with the overarching theme that the problem of an ascendant China / China in general / etc. is that China is too inward-looking, that "You'll never be Chinese." I know you'll never be Japanese, but I think China and its people and culture are quite different; I didn't spend as much time living in China or studying Chinese as the author, but in that time and in my experiences I think there is a lot more interest and openness of people (and many elites) than is given credit.

Unfortunately, there are huge structural and institutional barriers too...


I am tremendously excited to hear about this project; it's exactly what I want(ed) to do, these billionaires have just gotten the jump on me by a few (dozen) zeroes of net worth.

If any of them are reading, or anyone here knows them, I would do anything to get involved in this effort; the development and exploration of space is the most important mission in human history, and I'd love to put to work my business expertise, youthful energy, and eagerness to contribute... I'm well credentialed but more importantly extremely passionate!


> [...] I would do anything to get involved in this effort; [...]

Have you applied for a job at SpaceX, yet? That's one way to get involved with modern space faring.


It's an interesting proposition, but my intention has always been to find where I can be of greatest leverage. SpaceX is at a place in time / life / hiring that I think I wouldn't really be able to contribute to its mission (and mine) as much as I might by doing something else first. That plus I doubt I'd be of terrible appeal. The following are also true:

a) SpaceX appears more specifically to be looking for rocket engineers, etc, a background I do not have. (Good business / analytics / etc, but, a rocket engineer I am not). This problem likely applies to Planetary Resources as well. However...

b) Planetary Resources is more directly aligned with the mission I find most important - SpaceX is doing great things, and Elon Musk is super cool for insisting on his goal of making humans a multi-planetary species, but that's still a ways out. I don't know that I'd have as much ability to contribute. It's for this reason I've been trying to gain skills that would be of greater value down the line (business ones, in particular) in order to do what's necessary to help run and grow a commercial space venture.

c) I actually asked Elon Musk something on the subject - 'What advice do you have for people interested in getting involved early in these kinds of capital-intensive but world-changing businesses' and his answer was that, of course, it's helpful to have a few $100MMs from something like an internet company first :)


Indeed. I guess I just took the "I'd do everything." a bit too literal.


60-65 hrs a week isn't so many :)


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: