The most probably IBM CEO Ginni Rometti is referring to the development of their project Debator - AI answers questions. Thus collecting all the information on an employee IBM is able to ask "Do you want to quit?" Unfortunately it works...
Amazon woes and wins, IBM thinks it's solved employee happiness and Duplex phony phone calls everywhere!
Hello, here's a quick lowdown on what's been happening in the machine learning this week. Shareholders want a piece of Amazon Rekognition: Amazon investors will be allowed to vote if the company should be allowed to sell its facial recognition software to the US government, and if the technology should be subjected to an …
COMMENTS
-
-
-
Sunday 7th April 2019 12:16 GMT I.Geller
Re: The answer to the question "Do you want to quit?"
No, not at all! IBM AI searches using subtexts and subtexts. That is it really understands what is said and therefore can detect what was not there, with some probability. IBM calibrates all these probabilities and finds the "traitors".
What the wonderful time we are living in!
-
-
Monday 8th April 2019 16:34 GMT Anonymous Coward
@Mage - Re: it really understands what is said
Doesn't matter much. It is more important to convince those who are buying that machines are better than humans at understanding humans. We don't need perfect AI, just good enough AI that will be marketed as perfect AI. Unfortunately this works so well that I'm worrying for the future of the humanity.
-
Monday 8th April 2019 16:48 GMT I.Geller
Re: it really understands what is said
You are tragically mistaken... Each word of any text has its explicit context (in which it is used in this text), and subtext (which implicitly says what it means). For example, a dictionary definition is its word's subtext.
Furthermore, almost all words refer implicitly to other texts, which are similar to the given by their contexts and subtexts. So, AI can easily discover both dictionary definitions and other texts, know what you implicitly think but don't say. That's what IBM uses, determining whether you want to escape to freedom.
-
-
-
-
Sunday 7th April 2019 12:52 GMT Teiwaz
Well that's a little obvious
Possibly it just spews positive IBM cheerleader ra ra at the user and then measures the returning waves of cringe and cynical hardening like radar.
Could also measure the drying out of the spirit like drying and cracking wood as staff cross the threshold on monday morning.
-
Sunday 7th April 2019 16:22 GMT macjules
Watson: Are you happy at IBM?
Answer: Yes
Watson: I don't believe you. For the last time ARE YOU HAPPY AT IBM?
Answer: Yes, but I would like a bigger helicopter.
Watson: Ok Ms Rometty. I assess that there is a 96% chance of your leaving unless we give you a huge bonus and a bigger helicopter.
-
Sunday 7th April 2019 21:47 GMT Anonymous Coward
I know someone who used to work at IBM and from stories they've told then 20-ish years ago the key inidcator that someone was leaving IBM was when a salesman went round all the systems engineers to get info on the potential for sales to all their clients to "build up a strategy to maximize revenue over the next couple of years" as it turned out that that was a clear sign that the salesman had just landed a job at Amdahl!
-
Monday 8th April 2019 10:59 GMT Anonymous Coward
"We have asked IBM for more details."
My Guess is, they're actually using data taken from tools that were intended for security monitoring.. Hence why they're reluctant to talk about it.
I'm not talking about the IPS/IDS, no sir, the darker arts stuff, the tools that companies don't like to talk about - HTTP access requests, Email history and other more intrusive tools used for "Insider Threat Monitoring", such as ObserveIT and Teramind (Which are capable of collecting huge amounts of meta-data about user-interactions on their PC, including file accesses, email content, browsing patterns, time spent in applications etc etc ***)
With these, they can create employee profiles and identify patterns of leavers.
*** On a side note: I've known for some companies to use such tools to monitor an employees "productivity".
-
-
-
-
Monday 8th April 2019 06:26 GMT Anonymous Coward
Re: LinkedIn Profile
ROFL...
Mine was created and somehow never updated even though I'd moved jobs several times. I just never got around to logging in and working on it.
Like most people here who have (or had) one, it seemed like a good idea at the time.
Now? Nah. Why post stuff about you that can help identity thieves eh? Giving people access to who and where you work are nise snippets of data that those who want to suck your finances dry a nice big helping hand.
I deleted or rather attemped to delete mine years ago but LinkedIn and now MS still send me emails about some new feature and how cool it would be to link all my MS accounts together etc etc etc.
Sigh. Time for a new spam filter I think.
-
-
Monday 8th April 2019 08:26 GMT DrXym
Re: About to quit?
I expect it consists of little more than tracking a person's LinkedIn profile and their email / web activity. As IBM appears to want their workers to quit, they probably use the tool as a measure of success, not failure.
I wouldn't be surprised either if LinkedIn sells web service APIs that make it easy for big employers to build apps that snoop on their employees, e.g. periodically checking if a person's profile has been updated, or they've started linking to recruitment agencies recently.
-
Monday 8th April 2019 13:39 GMT rdhood
Re: About to quit?
:I wouldn't be surprised either if LinkedIn sells web service APIs that make it easy for big employers to build apps that snoop on their employees, e.g. periodically checking if a person's profile has been updated, or they've started linking to recruitment agencies recently."
I am guessing that they do this, and other stuff like it.
-
-
-
Sunday 7th April 2019 15:46 GMT The Man Who Fell To Earth
How sophisticated can the IBM AI really be?
IBM AI: How old are you?
IBM Employee: 52
IBM AI: 52>40.
IBM AI: Therefore there is a (52-40)/(64-40)x100% chance you will
be firedleave in the next year.*-
-
Monday 8th April 2019 13:46 GMT I.Geller
Re: How sophisticated can the IBM AI really be?
IBM relies on the analysis of both contexts and subtexts. That is, IBM sees not only what you said publicly, but also what you think covertly - your choice of subtexts helps IBM.
Dictionary definitions are one example of subtext, they convey your general mood-emotions, testify if you are the traitor ans want to leave IBM. Clear?
-
-
Monday 8th April 2019 19:12 GMT GrapeBunch
Re: How sophisticated can the IBM AI really be?
Mine contained no arithmetic at all.
Do you work for IBM? If "Yes" then no, you won't quit in the next week.
I feel confident that will be accurate more than 95% of the time.
Mine's the one with the vocal override that doesn't understand numbers, but always gives the best answer.
-
-
Sunday 7th April 2019 16:04 GMT Mike 16
Valuable planning info.
If a manager can be confident that an employee is likely to leave in the near future (barring any effort to address the issues), and can speed that future's arrival with a little subtle harassment, the company can save on both severance pay and HR paperwork they would otherwise spend on firing them. And that will look good on the manager's "social credit" equivalent. The damage that either course of action would have on future product quality/sales (compared to addressing the issues) will show up later, when the execs have already sold what shares they could during the bump from a RIF. And they will have done a round of musical chairs such that no blame for this can be solidly traced back to them. WIN/WIN for execs and managers, LOSE/LOSE/LOSE for workers, customers, and future shareholders, but "Not my job, man"
-
-
Monday 8th April 2019 21:57 GMT fidodogbreath
Re: Alexa and sensitive patient data
"Protected" health data is far more valuable to credit bureaus and their clients, big pharma, pharmacy chains, medical-equipment sellers (hmmm, Amazon sells some of that...), etc.
It's only a matter of time before an unfavorable diagnosis will result in your credit cards being cancelled before you get home from the doctor's office, for just one one dystopian example.
-
Monday 8th April 2019 22:22 GMT martinusher
Re: Alexa and sensitive patient data
HIPPA means that while your data might be handled by Alexa (i.e. AWS) it would effectively be in a tunnel between you and your provider. Amazon won't touch this data because both the legal penalties and the optics if they got found out would be very undesirable. Amazon has plenty of other places to make their cut; for example, they will be in the catbird seat when it comes to supplying goods and services to both user and provider.
Although the legacy health system in the US is for profit fee for service, a setup that encourages upselling of the most expensive services and drugs, in reality health systems find it more cost effective to use a universal health care model like the NHS. This puts a lot of emphasis on preventative and continuing care. A peripheral like a voice interface could be a very cost effective substitute for round the clock nursing care, something that's not practical or desirable for most people. It can not only issue reminders but interact with a person who has a problem either explicitly ("Help, I've fallen" etc.) or implicitly (failure to detect routine activities, hear unexpected noises etc.). The potential of this setup is immense.
-
Tuesday 9th April 2019 02:56 GMT fidodogbreath
Re: Alexa and sensitive patient data
HIPAA means that while your data might be handled by Alexa (i.e. AWS) it would effectively be in a tunnel between you and your provider.
Unless the TOS includes a HIPAA waiver.
"Defendant moves for summary dismissal, Your Honor. The HIPAA waiver is clearly stated on page 38 of the Terms Of Service. At [exact date and time], the plaintiff asserted that he had read and agreed to this binding contract in its entirety. On page 47, the plaintiff further agreed to resolve all disputes by binding arbitration only, using arbitrators that are selected and paid for by the defendant."
-
-
-
-
Monday 8th April 2019 02:51 GMT Michael Hoffmann
Alexa Skills Kit
"The software kit, dubbed Alexa Skills Kit, has only been shared with a few select organisations so far, based on an “invite-only program”.'
Uhm, no. ASK is the "standard" SDK/API available to all and sundry who are signed up for Alexa development. Takes 5 minutes to sign up and is hardly "invite only".
I presume there is a separate SDK as addon to or within the ASK that has these health additions?
-
Monday 8th April 2019 06:32 GMT Steve Davies 3
About to leave for another Job?
If you are at IBM, you will be hoping to get canned the day after you have your new job offer in your pocket.
Otherwise it should be BAU until the day to tell your boss that you are leaving. Give the PHB no warning at all and you might get that raise you want but be prepared to walk (or be escorted by security) out the door within the hour of giving notice. Garden Leave is very welcome at this time of year for... Gardening. :)
-
Monday 8th April 2019 13:44 GMT Anonymous Coward
Re: About to leave for another Job?
As a contractor at IBM (as in, even *LESS* job security than being a regular minion), the ideal is to give IBM **exactly** as much advance notice as they'd give you (1-2 hours during the weekdays/work days, none if it's a non-working day).
Then the time they decided to make a 10% cut of how much they were paying for your "purchase order" (your contract) which the contract company passes on to you. My comment was "well, for 10% less money, I can just make 10% less effort".
So yes, the formula should be easy enough for even Watson to figure out:
IF (IBM_employee)
then (chance_of_quitting: = 100%)
FI
(hey Reg: how do you format and indent code on this site?)
-
-
Monday 8th April 2019 08:49 GMT StuartK
Game it
Knowing your every move is being judged by an AI is going to lead to a lot of second guessing and probably an increase in computer doesn't lie related workplace stress.
Also, the knowledge the algorithm exists would seam likely to modify it's accuracy, as there will now be a tempation to game it (keyword combinations in emails, holiday patterns, clock in/clock out trends, performance review metrics) to either initiate some kind of offer from your employer to keep you around, or at least find out how much you are worth keeping around if there's a deafening silence. And maybe, if you find the criteria and subvert expectations, the algorithm gets less accurate.
-
Monday 8th April 2019 16:02 GMT Anonymous Coward
Re: Game it
Also, the knowledge the algorithm exists would seam likely to modify it's accuracy, as there will now be a tempation to game it (keyword combinations in emails, holiday patterns, clock in/clock out trends, performance review metrics) to either initiate some kind of offer from your employer to keep you around,
This is IBM. They don't want to keep *ANYONE* around.
-
Monday 8th April 2019 16:20 GMT I.Geller
Re: Game it
That's not possible, my dear.
AI is a terribly insightful thing, the chance to cheat on it very small. AI "feels" what you implicitly think but do not explicitly say, analyzing subtexts of your contexts. Dictionary definitions are, for example, these subtexts: the more layers you use forming them the more they tell about your subconsciousness, and the more AI knows about you.
-
-
-
Monday 8th April 2019 13:20 GMT I.Geller
Re: Alexa can talk to you about your health privately
It can, Amazon annotates/ trains patterns
- by dictionary definitions
- and all other a patient's private texts, mined from his computer,
- forming the patterns' explanations/ tuples,
- as the result Alexa really understands what a patient needs/ doesn't need to pry and spy and steal on-line. This is AI database off-line technology and Alexa can talk privetly.
-
Monday 8th April 2019 22:14 GMT FozzyBear
Re: Alexa can talk to you about your health privately
And yet the shareholders have taken steps through the SEC to try and stop the facial recognition software due to possible privacy and civil rights. Completely ignoring, Alexa, the most blatant abuser of privacy.
One might be forgiven thinking this is just a political move by some special interest groups. Unfortunately gross stupidity can't be discounted
-
-
Monday 8th April 2019 09:56 GMT Mage
Duplex: Google doesn’t seem bothered
Well, an advertising company harvesting private information wouldn't.
"and has instead allowed people to opt out of receiving calls from Duplex instead."
So WRONG. The default should be opted out, people should have to opt in. How on earth would opting out work anyway? You'd have have to give Google info, log in and do that on an Google account. Ah, that's why it's allowed by default.
Reminds me of USA CAN-SPAM and the fact that USA Robo callers and cold callers might sometimes get fined but little is collected.
This will also be a boon to those "offering support for windows" via phone. They won't have to employ probably unsuspecting Indian Call centre staff. The last one of those I got had a spoofed Caller ID in Iceland!
-
-
Monday 8th April 2019 18:57 GMT I.Geller
Re: Deeply misunderstood
The programming is dead as a profession, and all programmers will be unemployed in the next 5-7 years. Really, what are they doing? Manually structure texts into commands. What does AI do? Structures the same automatically into patterns. But the patterns are the direct analogue of the commands!
-
-
Monday 8th April 2019 21:15 GMT fidodogbreath
Quote-unquote
This week Amazon also released a toolkit for developers to allow its digital assistant Alexa to access and transmit sensitive patient data "privately" for healthcare companies.
Forgot the quotes around "privately."
When a tech company claims that something is private, or deleted, or covered by a tracking option that users naively think they can turn off, that just means that said data will no longer be surfaced in the customer-facing part of the UI.
Everything that passes within view of the panopticon is stored and monetized forever, and there is exactly jack shit that any of us can do about it.