Slashdot Mirror

← Back to Stories (view on slashdot.org)

Executives Say AI Will Change Business, But Aren't Doing Much About It (axios.com)

Posted by msmash on from the how-we-are-handling-it dept.

American business executives expect artificial intelligence to have a large impact on their companies, but few are actually doing anything with AI, according to a new MIT- Boston Consulting Group survey. From a report: Key takeaways, per co-author and BCG senior partner Martin Reeves: Nearly 85% of the 3,000-plus executives surveyed expect AI will give them a competitive advantage But their adoption of AI isn't matching up: just 1 in 5 of the companies use AI in some way, and only 1 in 20 incorporate it extensively. "Less than 39% of all companies have an AI strategy in place," they wrote. The barriers for adoption include: access to data to train algorithms, an understanding of benefits to their business, a shortage of talent, competing investment priorities, security concerns, and a lack of support among leaders.

41 of 76 comments (clear)

  1. Re:Really, "Nut"? by kwerle · · Score: 2

    Clearly it has not affected the use of spell or grammar checkers.

  2. At least 35.03% are wrong. by xxxJonBoyxxx · · Score: 1

    >> Nearly 85% of the 3,000-plus executives surveyed expect AI will give them a competitive advantage

    I am quite certain that at least 35.03% of them are wrong.

  3. Sure, when it happens by avandesande · · Score: 1, Flamebait

    Pattern recognition is not 'AI'. Maybe when some actual AI is developed companies will adjust to it.

    --
    love is just extroverted narcissism
    1. Re:Sure, when it happens by bluefoxlucid · · Score: 1

      "They say a future technology will make them more-competitive but aren't using cold fusion RIGHT NOW!"

      --
      Support my political activism on Patreon.
    2. Re:Sure, when it happens by avandesande · · Score: 1

      Web 2.0 has run its course and IOT is stillborn they are grasping at straws.

      --
      love is just extroverted narcissism
    3. Re:Sure, when it happens by ShanghaiBill · · Score: 3, Insightful

      Pattern recognition is not 'AI'.

      Pattern recognition is what your brain does. So how is it "not AI" when a computer does it?

    4. Re:Sure, when it happens by MightyMartian · · Score: 1

      Part of the issue here is that there's the popular definition of "Artificial Intelligence", which seems to be what the marketers and technology journalist have latched on to, and then there's AI research. Obviously, as with any area of science or engineering, the researchers are going to be using a far more rigorous definition.

      But really I think what's being talked about here, for instance, in replacing or augmenting articled clerks with AI searches is a bit more elaborate than regular expression searches. This makes this sort of search, whether you call it AI or pattern matching or something else, fairly ambitious. In the long run, I suspect the popular and technical definitions of AI will ultimately merge; you will have systems that have some degree of intelligence.

      --
      The world's burning. Moped Jesus spotted on I50. Details at 11.
    5. Re:Sure, when it happens by MightyMartian · · Score: 1

      That's certainly an important function of the central nervous system of all animals, but our brains do a lot more than just sift data.

      --
      The world's burning. Moped Jesus spotted on I50. Details at 11.
    6. Re:Sure, when it happens by avandesande · · Score: 1

      Sure, it's more elaborate but it's just an algorithm. Nobody thought Google was AI when they released their search engine and it was pretty. Because this is mostly an underhanded marketing tactic aimed at buzzword compliant PHBs it is precisely the reason why companies should avoid a vendor using the term AI and stick with real solutions.

      --
      love is just extroverted narcissism
    7. Re:Sure, when it happens by zlives · · Score: 1

      is it too late to welcome our robotic AI overlords? because by "long run" you mean Christmas shopping time right?

    8. Re:Sure, when it happens by AHuxley · · Score: 1

      Any sufficiently advanced pattern recognition is indistinguishable from a real AI.

      --
      Domestic spying is now "Benign Information Gathering"
    9. Re:Sure, when it happens by bluefoxlucid · · Score: 1

      IoT is getting a lot of bad publicity for being an enormous security nightmare. It needs a standard. Not another standard so we have 14 standards; it needs a standard that people follow. As for regulations... Congress should stop at accountability for reasonable security measures; legislating technology creates inflexibility.

      I've left mine stillborn, though. I wanted IoT devices to have a near-process set-up (i.e. you have to put the devices together, tell them they're setting up, and they open a window to communicate with one device) to exchange keys between an IoT hub and either an IoT device or client. Yes, I said either: your IoT device does not talk to clients because I don't want it managing its own security.

      The whole purpose is to have the IoT device access-controlled by the IoT hub, listening to no other connections. Today, you plug an IoT device in and you can hit Telnet, a Web server, the like. The devices find each other, network together, and talk to a service in the cloud. I don't want them accepting any connections; instead, you connect to the hub, and it proxies the connection.

      The IoT hub would use the small negotiation window (activated by a hardware button) to establish a credential--it and the client exchange public keys either internally-generated, from U2F (a Yubikey), or some other mechanism. They trust each other by the identity of that key. When a device--your phone, computer (U2F), etc.--connects to the IoT hub outside the negotiation window, the IoT hub checks for certificate validity. That means the only code run on outside, unauthenticated connections is the kernel's network stack, the Web server's connection handling, and the TLS library's certificate validation. If you don't have a first-factor U2F trust established with anything, then it flatly rejects any U2F authentication; otherwise it also runs through some U2F code, and validates the certificate through the same TLS library code.

      Until you're authenticated in that way--which is user-transparent--the attack surface is tiny. It's possible to have a miniature Web server handle these new connections and then sendmsg() them to nginx, handing off a completed TCP connection with a fully-authenticated client TLS certificate; you'd have to write the module to transfer the relevant connection state information (session key, etc.) to nginx, but POSIX does allow you to give an established TCP connection to another process.

      Now your devices are secure.

      The other part is to run services inside Docker containers. NestCam should be able to run its service on an IoT server, which connects to the IoT hub. The IoT hub can allow communication between NestCam and the NestCam local service. That can communicate with a storage back-end (FreeNAS, SMB, FTP, whatever). Further, when you pair your phone (e.g. via Bluetooth) to the IoT hub and authenticate the Nest app, the Nest app would be able to retrieve the IoT Hub's IPv6 address, thus becoming capable of accessing your Nest cameras from anywhere in the world--through a TLS-authenticated, encrypted connection, directly to your house, passing through the IoT Hub, using only the local service on your IoT Server.

      You don't even need a certificate authority. You physically walked up to a device in your house, pushed a button, paired with it over Bluetooth, and carried out a key exchange. You have a direct trust.

      Sadly, I don't have the patience or depth of knowledge to produce a full specification or reference implementation.

      --
      Support my political activism on Patreon.
  4. What is AI? by Volda · · Score: 3, Interesting

    Can someone give me the definition of AI? I ask because it keeps getting thrown around but when people get in to details they talk about pattern recognition, machine learning or data analytics. Not what I would consider AI. CTO seems to think that if a machine can read, say safety manuals, then it can make decisions on safety better then humans. We have yet to see this work.

    1. Re:What is AI? by Anonymous Coward · · Score: 1

      (With tongue firmly planted in cheek) AI is the set of problems computer researchers think computers could do, but can't yet.

      Once there's off-the-shelf software to solve a category of problem, it stops being "AI" and becomes "machine vision" or "autonomous $FOO".

    2. Re:What is AI? by clodney · · Score: 4, Interesting

      So long as AI is implementing techniques that work on general purpose computers, programmers will look at it and say that that is not AI, that is just code running this algorithm or that. You keep waiting for the magic to happen, and keep finding out that it is just software.

      But so what? Neural nets are making great strides in specific applications, and even though we know how they work in general, the specific way they put together associations still surprises us and lets them come up with answers we didn't expect, or implementations we would never have come up with. Computers playing Jeopardy don't do anything a human can't do, but the fact that they can do it at all is a huge leap forward compared to where we were 20 years ago. And sure the hardware is a million times faster, but as the saying goes, quantity has a quality all its own.

      Going back to the point of magic happening, what would it take for you to decide something was AI? And if you discovered you understood all of the techniques that went in to that, would it stop being AI?

    3. Re:What is AI? by Rick+Schumann · · Score: 1

      The waters are so muddied by marketing hype, media hype, and hype from so many other sources, that it's easier to explain what is not 'AI': Nothing that you've seen or heard about is 'artificial intelligence'. It's all ersatz. At the rate we're going we won't have real 'AI' anytime soon, probably not in our lifetimes -- not unless neuroscience makes a big breakthrough in how our own human minds are able to actually 'think' and be 'conscious' and 'self aware' like we are; in other words we have to solve the riddle of "I think, therefore I am" before we can create machines that can do the same thing.

    4. Re:What is AI? by gravewax · · Score: 1

      At this point there is no real definition. Various companies, researchers and media have made such a fucking mess of the messenging from this space that anything that in anyway resembles a computer providing an answer to something a human would normally do is considered fucking AI. Even if said program is just a simple series of IF THEN statements that produce pre programmed speech some retard will call it AI.

    5. Re:What is AI? by Myrdos · · Score: 1

      That's why I prefer the term "Synthetic Artificial Virtual Intelligence". So much less debate about the meaning, letting us focus on the practical benefits.

    6. Re:What is AI? by speedplane · · Score: 1

      The marketing hype is itself changing the definition of AI. Under the newer definition it means a computer doing something that appears to be intelligent. Under this definition, AI is all around us.

      --
      Fast Federal Court and I.T.C. updates
    7. Re:What is AI? by Myrdos · · Score: 1

      Why does it need to be exactly as smart as a human in order to be AI? I mean, cats aren't self aware. Why not as smart as a cat?

    8. Re:What is AI? by avandesande · · Score: 2

      You admit then it's just more of the same of what we have experienced the last 25 years. Why is it now suddenly AI? It's just a marketing gimmick.

      --
      love is just extroverted narcissism
    9. Re:What is AI? by avandesande · · Score: 1

      Okay, well if someone makes a robot as smart as a cat I am in.....

      --
      love is just extroverted narcissism
    10. Re:What is AI? by Tablizer · · Score: 1

      I'll call it AI when it can fetch me a beer, take out the garbage, and suck my dick.

      --
      Table-ized A.I.
    11. Re:What is AI? by taustin · · Score: 1

      AI is a marketing term for vaporware. It's an important tool to sell stock options to investors.

    12. Re:What is AI? by Rick+Schumann · · Score: 1

      I did not say, "as smart as a human", that is something YOU are superimposing on what I did say, which is being able to 'think' like a human brain can.
      Also your cat or dog is, in fact, smarter than the junk they're trying to pass off as 'AI'.

    13. Re:What is AI? by Rick+Schumann · · Score: 1

      Since you bring it up.. at current we can't even build a machine that emulates the cognitive abilities of a cat's brain, either; they don't even really know how an animals brain works yet.

    14. Re:What is AI? by Graydyn+Young · · Score: 1

      The definition of AI has changed. We all need to get used to this. If you throw a fit every time somebody says "AI" without meaning some kind of magical super machine, all of your hair is going to fall out. The meanings of words change over time. Just go with it.

      The modern definition is basically just any implementation of machine learning. Which is funny, because the phrase "machine learning" also used to be a buzzword that nobody in the industry actually used unless talking to the media. We'd be more likely to say "Computational Statistics" or "Function Estimation". So, in summary, the modern definition of AI is: statistics on a computer.

    15. Re:What is AI? by clodney · · Score: 1

      You admit then it's just more of the same of what we have experienced the last 25 years. Why is it now suddenly AI? It's just a marketing gimmick.

      The techniques were described as AI 25 years ago as well, it is just that the scale and applicability of them is hugely different now.

      Perhaps where we are foundering is expectations. When I took an AI course back in college we learned about expert systems, neural networks, associative arrays and other things I don't remember, and all of those were algorithms or approaches that would be used to build AI. I have no problem with calling useful systems built on those techniques AI, even if they don't pass the Turing test.

    16. Re:What is AI? by myrdos2 · · Score: 1

      Yes, I follow. But why does it need to be able to think like a HUMAN brain? Lots of animals have some intelligence without being self aware.

    17. Re:What is AI? by avandesande · · Score: 1

      It's happened through incremental improvements. So what happened in the last year or two where suddenly we need AI consultants that wouldn't be covered by ordinary IT initiatives?

      --
      love is just extroverted narcissism
    18. Re:What is AI? by Rick+Schumann · · Score: 1

      We're trending towards so-called 'AI' being put in charge of critical things, some of which involve human safety, the most notable of which are so-called 'self driving cars'. The garbage they're trying to pass off as real 'AI' doesn't know the difference between a human being and any other inanimate object; that's because it cannot actually THINK.

    19. Re:What is AI? by Rick+Schumann · · Score: 1

      Even a dog or a cat can think, even if it's on the level of about a 2-year old human, and that's still better than the so-called 'AI' they keep trotting out.

  5. Still primarily in R&D by clodney · · Score: 1

    I think the reason few are actually doing anything with AI is that it hasn't been turned into a product yet. Ask a financial analyst how they intend to use AI to improve their forecasts, and they will give you a blank look. Sell them a product that you feed a bunch of data into and a forecast spits out, that uses AI under the hood, and they will be happy to buy it. But they don't have the ability to start from scratch.

    Kind of like a study asking "Why aren't house builders using superconductors?" Because they have no idea how to put them to use until they get out of lab settings and get built into products.

    1. Re:Still primarily in R&D by avandesande · · Score: 1

      A few years ago nobody claimed speech recognition was AI. IBM started this silliness when it started a marketing campaign code named 'Watson' and now I guess everyone should hire consultants to find out how to get some 'AI'.

      --
      love is just extroverted narcissism
  6. Re:AI not existing makes planning for it hard by ShanghaiBill · · Score: 1

    Hard AI does not exist.

    AI does not need to be hard/strong in order to be useful to businesses.

  7. product support by Tablizer · · Score: 1

    I suspect the product support lines will be the first to use it, not because it's good, but because co's want to cut staff. It just has to kinda sorta work to make it tempting enough. It's one of the lowest barriers of entry due to low expectations since product support already sucks at a good many co's.

    --
    Table-ized A.I.
  8. LipFlap by sdinfoserv · · Score: 1

    Since apparently nobody really knows what "AI" is:
    https://qz.com/1067123/stop-pr...

    saying an undefined quantity will accomplish something is a bit of a stretch.

  9. Overhyped by manu0601 · · Score: 1

    AI joined the recent overhyped buzzwords: MOOC, 3DTV, AR/VR... When the dust will settle, a few will certainly use it, but we will hear much less about it.

  10. In other news by coofercat · · Score: 1

    In other news, surveys show that 85% of executives have no imagination and the attention-span of a gnat. They have no interest in what's happening next quarter, let alone what might happen in a year or two.

  11. American business executives regurgutate by mark_reh · · Score: 1

    whatever BS is included in the latest best-seller business book summaries they read, or whatever they read in that airline magazine on their last flight (you know, the article that was sandwiched between Sharper Image ads for electrically heated dog sweaters and the ad for $500 per person steak dinners). They'll do something with it when one of their hired-gun management consultants tells them what they should do with it.

  12. What is AI? by bdwoolman · · Score: 1

    Good question. And question is the operative word. First IMHO true AI should be indistinguishable from I. Einstein thought computers were uninteresting because they did not ask novel questions. They still don't. An AI should be able to synthesize data sets and generalize across them, posit questions and set novel goals and elucidate tasks to reach them. An AI should be able to see what isn't and ask , "Why not?"

    Currently, I see AI as just a marketing term for highly capable systems that can perform tasks that people have performed historically. I am not being flippant when I say that, though many of these tasks are performed by people who possess intelligence, the tasks themselves require little of it. (Or none of it.) Not to say that the tasks are simple or that the do not require high levels of skill and ability.

    A deterministic system will perform a task as assigned. An artificial intelligence will say, "You know, boss, I have evaluated your request and I think there is a faster way to achieve the goal you defined. Shall I proceed as requested or do you wish me to elaborate on my observation? Coffee?"

    --
    "No fear. No envy. No meanness." Liam Clancy