681–690 of 711 posts
Earlier quoted context omitted.
Do you have a link to the commercial?
Yikes, I misremembered, it’s a Hewlett-Packard commercial. https://youtu.be/0EL3ZkcMNS4 This commercial still makes me laugh. The way she finishes his sentence with “our digital transformation” is the kicker.
Earlier quoted context omitted.
Machine learning could beat my doctor right now. "Using Random Forest Algorithm for Breast Cancer Diagnosis" https://ieeexplore.ieee.org/document/8644835 Most of what the doctor is doing is a type of modern shamanism though. Person doesn't feel good, so the doctor orders a useless test, test comes back negative so the person feels better. Then we complain health care cost too much. 10 full blood panel samples a year…
> Person doesn't feel good, so the doctor orders a useless test, test comes back negative so the person feels better. Person doesn't feel good. Mild flu-like symptoms. There's a good chance that the patient will get better if I do literally nothing. However, I also know that it could be X, Y, Z... So I order a test to prove that it's not those diseases because if it turns out to be them it would make me guilty of gro…
Also, because we haven’t done this at scale, we don’t know the true baseline for most people who may have something on the blood panel samples but never get tested because they are asymptomatic. So then we have this whole group of people with recommended treatment because of the results, even though they are asymptomatic. And they are treated, which then leads to increased costs and side effects…
And now we’ve effectively increased medical costs and decreased quality of life for asymptomatic patients. BUT - if we do it with just a drop of blood, we might be able to start a startup and raise some funding… We could call it Thermos or something…
Earlier quoted context omitted.
Funny, similar thing happened to me. IBM along a few other behemots pitched for a serious project at a company I worked for as enterprise arch. All companies brought their top salespeople, and all tried nasty things, but IBM was by far the worst. Their top guy started their pitch by saying he chatted with our CEO over the christmas holidays. He mentioned - and I am not making this up - that he should be talking to pe…
It sure sounds like IBM were bad. But it sounds like your company was even worse, the way your bosses' boss behaved.
Earlier quoted context omitted.
Deep Blue isn't ML at all, it's just a purpose built system for solving chess. It doesn't learn anything as such and just a fancy brute forcing machine that (smartly) goes through all possible chess moves and selects the best possible tree of outcomes. The work on actual ML happened independently of this.
That is a form of machine learning. It uses an algorithm with an enormous ability to look ahead, and selects the moves according to a heuristic that have been shown previously to be most likely to create the desired end state. Chess is not a solved game from the starting state, so it has to make assumptions based on the data it has. This is machine learning. Please don't make such definite statements. You even say it…
Unless I’m misinformed, this part isn’t true. The heuristic was hardcoded with the help of human experts.
Earlier quoted context omitted.
I don't think NLP is the problem here. Even if you crack NLP perfectly and would somehow be able to build a semantic map, or whatever your technology would use, of the user request with perfect fidelity, there's still the question of actually understanding what it's about. Perfect NLP would help in very limited number of situation where the meaning is clear - i.e. doing some pre-defined action, like closing an accoun…
> there's still the question of actually understanding what it's about. Philosophically interesting, but in practice entirely unimportant. What you want is something that speaks like a human would speak in a similar environment, then the issue (getting the correct responses) is solved. Whether or not it actually understands won't change a thing about that situation.
Earlier quoted context omitted.
I remember reading some very amusing Chef Watson cocktail recipes. Lots of savory concepts like putting chicken or meat into a drink.
I worked on Chef Watson and was part of the team at SXSW. One of the most amusing parts was how much it loved bacon. This was because bacon could be sweet, savory, salty, and fatty - it all depended on how you cooked it. A big part of the way that Chef Watson created recipes was by connecting flavors together. There was a while when it didn’t record why it had included an ingredient - so bacon would be brought origin…
Earlier quoted context omitted.
Deep Blue preceded this. They had specialty equipment for crunching chess moves. Really excellent hype machine. I liked the documentary about it: https://m.imdb.com/title/tt0379296/
Was Deep Blue/Watson basically the starting point of AI/ML, or had that already been happening earlier or concurrently?
As someone doing a CS degree now, I seem to be the only one who doesn't want to have anything on my resume to do with "AI", blockchain, ML, NFT, chatbots, etc... all I see is overhyped product after product, one-size-fits all solutions that frustrate customers and create problems for humans to clean up, hugely valued companies that have very little real improvement over conventional technology, etc. An "AI chatbot" i…
On the cynical flipside though, you can't entirely keep "buzzwords" off your resume simply because recruiters will always squint and "find them" because there's money involved if they do. You are almost always going to get recruiters for whatever the day's buzzword is.
I have a Masters degree and Python experience, and to most AI/ML recruiters that looks like "possible AI/ML researcher". They aren't entirely wrong, I did study AI/ML in grad school. They aren't very right either, because I learned enough to be "dangerous" and came out of grad school a massive AI/ML cynic. (That the field is mostly just Sparkling Statistics and people in general are bad at statistics and easily wowed by Sparkling Statistics. That this isn't the first hype cycle in the field, and it isn't likely to be the last either; we've not really improved on the research of the AI boom in the 1960s/1970s nor the other big AI boom of the late 1980s. They had chat bots then, they had most of the algorithms figured out, including their weaknesses. This boom we've just massively increased Garbage In, Garbage Out to those algorithms and given ourselves enough blinders to pretend that everything is still under control. We haven't solved their weaknesses. We haven't actually made major conceptual leaps. We just have a lot more data and a lot more speed. Which is something the 1960s researchers both predicted and warned us about.)
So yeah, your cynicism here is very valid, at least in my opinion. There's not a lot you can do about it, especially if your aim is to outright avoid recruiters chasing you for whatever VC money is getting thrown at them. I don't have a lot of other advice here other than the only thing you can really do with such cynicism is to apply it as best as you can to improving the things that you have the power to improve.
Ironically, the main project I work on today involves a lot of AI/ML and I'm known as a bit of a wet blanket on the team trying to temper expectations and trying to keep us from doing the worst mistakes ("confidence" values do not mean what people think they mean and showing them in any UI anywhere, especially under the name "confidence" is a massive mistake; machines don't have the "ego" for "confidence", it's a poorly applied statistical term of art that people use to badly anthropomorphize the statistical models). I don't win every battle and I do my best to make it clear I come from a point of pragmatism. It's a hard balance to keep.
Earlier quoted context omitted.
Good question. Prediction markets [1] allow people to bet on outcomes and benefit financially if they are correct. However, many jurisdictions ban them outright claiming that they a form of gambling and challenging the unregulated nature of questions leading to misaligned incentives (“When will that building burn down?”). Yet despite these issues, the scheme can offer a neutral ground for betting against overhyped te…
So all you're saying is that cryptocurrencies can be deployed to facilitate illegal behavior. We've known that for quite a while.