zenpundit.com » AI

Archive for the ‘AI’ Category

On the Intelligence of the Artificial, thus far

Saturday, August 12th, 2017

[ by Charles Cameron — YouTube removing videos that document war crimes in the Middle East, Elgar ]
.


YouTube removes video documenting Islamic State’s destruction of Nimrud artifacts (AFP)

**

Here we are on the merry-go-round:

YouTube AI deletes war crimes evidence as ‘extremist material’

YouTube is facing criticism after a new artificial intelligence program monitoring “extremist” content began flagging and removing masses of videos and blocking channels that document war crimes in the Middle East.

Middle East Eye, the monitoring organisation Airwars and the open-source investigations site Bellingcat are among a number of sites that have had videos removed for breaching YouTube’s Community Guidelines.

The removals began days after Google, which owns YouTube, trumpeted the arrival of an artificial intelligence program that it said could spot and flag “extremist” videos without human involvement.

But since then vast tracts of footage, including evidence used in the Chelsea Manning court case and videos documenting the destruction of ancient artifacts by Islamic State, have been flagged as “extremist” and deleted from its archives.

**

Without human involvement?

Humans were involved in the decision to let this AI proceed in real time “without human involvement”.

H Sap or AI? — who’s to trust?

**

YouTube, Nimrod Barenboim conducts Elgar:


Compare “YouTube removes video documenting Islamic State’s destruction of Nimrud artifacts”, illustrated above

Sunday surprise, from dogs & muffins to Stop! & 45mph limit

Sunday, August 6th, 2017

[ by Charles Cameron — better eat a muffin than a dog, stop at a stop sign than blow through it at 45, oh well ]
.

**

It’s cute, sorta, that AIs can’t easily distinguish dogs from muffins:

But STOP!

**

What’s not so funny is that the AI in many autonomous vehicles misreads a treated STOP sign —

— as a sign for a 45 mph speed limit. As Ivan Evtimov and colleagues indicate in Robust Physical-World Attacks on Machine Learning Models, “Physically realizing such an attack for road signs can raise concern in human observers.”

I’ll say.

Grab the wheel, conscious entity!

**

Readings:

  • Becoming Human: AI, Why are Marketers All Talking about AI Now?
  • Wired, Simple Pictures that State-of-the-Art AI Still Can’t Recognize
  • Car & Driver, Researchers Find a Malicious Way to Meddle with Autonomous Cars
  • Arxiv, Robust Physical-World Attacks on Machine Learning Models
  • Ouroboros catch’em post

    Tuesday, July 18th, 2017

    [ by Charles Cameron — attempting to keep the self-eating serpents in one pen, so they don’t get tangled in your hair and eyes ]
    .

    From Aneurism, a brilliant long-form essay by neurosurgeon Henry Marsh, from his book Do No Harm, and presented in Slightly More Than 100 Exceptional Works of Journalism:

    Are the thoughts that I am thinking as I look at this solid lump of fatty protein covered in blood vessels really made out of the same stuff? And the answer always comes back–they are–and the thought itself is too crazy, too incomprehensible, and I get on with the operation.

    From Political Tracts of Wordsworth, Coleridge and Shelley via PR Beckman:

    The purpose of the historian, to Coleridge, is the same as that of the poet : to convert a series of events, which constitute the straight-line of real or imagined history, into a whole, so that the series shall assume to our understanding “a circular motion — the snake with its tail in its mouth.”

    Here’s one more:

    And this I can’t resist — there’s hope for humankind!

    There will no doubt be others, which I’ll drop into the comment section. So you don’t need to be troubled by a new post every time I see one.

    Here’s what “not even people” sounds like

    Wednesday, June 7th, 2017

    [ by Charles Cameron — Eric Trump & voices in the uncanny valley ]
    .

    First, here’s Eric Trump saying the Democrats are “not even people”:

    Now, for an exact comparison, here’s what “not even people” sound like, when they speak those same words:

    Listen to the gaps, Eric, the lapses of emphasis.

    Artificial human voices will no doubt improve: at this point they’re easily discernible. They sound uncanny.

    Dems don’t sound like that.

    Would a democracy of artificial intelligences hold a variety of opinions?

    Friday, June 2nd, 2017

    [ by Charles Cameron — opening a conversation ]
    .

    I’m hoping to engage some of my friends and net acquaintances — Peter Rothman, John Robb, August Cole, Jamais Cascio, Monica Anderson, Chris Bateman, JM Berger, Tim Burke, Bryan Alexander, Howard Rheingold, Jon Lebkowsky and no doubt others — in a conversation on this topic, here at Zenpundit.

    Starting as of now: with encouragement to come — send posts to hipbonegamer@gmail.com, any length, fire at will!.

    On the face of it, AIs that are seeded with different databases will come to different conclusions, and thus the politics of the company of AIs, democratically assessed — ie one AI one vote — would be stacked in favor of the majority of kindred DBs from which the set was seeded. But is that all we can say? Imaginatively speaking, our topic is meant to arouse questions around both democracy and intelligence, artificial and oitherwise. and politics, we should remember, extends into warfare..

    **

    Two announcements I saw today triggered my wish to stir the AI pot: both had to do with AI and religion.

    The first had to do with an event that took place last month, May 2017:

    Artificial intelligence and religion
    Theos Newsletter, June 2017:

    Can a robot love? Should beings with artificial intelligence be granted rights? The rise of AI poses huge ethical and theological questions. Last month we welcomed John Wyatt and Beth Singler from the Faraday Institute to discuss these issues.

    Specifically:

    Advancements in Artificial Intelligence (AI) and robotics have been making the headlines for some time now. Articles in mainstream media and features in prime-time television keep pouring in. There is clearly a growing interest in humanoid robots and the varied issues raised by their interactions with humans.

    The popularity of films such as Ex Machina, Chappie, I-Robot and more recently Her reveal an awareness of the challenges hyper-intelligent machines are already beginning to pose to complex issues such as human identity, the meaning of empathy, love and care.

    How will more advanced, integrated technology shape the way we see our families, our societies – even ourselves?

    and one event next year:

    AI and Apocalypse
    Centre for the Critical Study of Apocalyptic and Millenarian Movements (CenSAMM)
    April 5 – 6, 2018. Inside the Big Top at the Panacea Charitable Trust gardens, Bedford, United Kingdom
    CenSAMM Symposia Series 2018 / www.censamm.org

    We invite papers from those working across disciplines to contribute to a two-day symposium on the subject of AI and Apocalypse.
    Abstracts are due by December 31, 2017.

    Recently ‘AlphaGo’, a Google/Deepmind programme, defeated the two most elite players at the Chinese game ‘Go’. These victories were, by current understandings of AI, a vast leap forward towards a future that could contain human-like technological entities, technology-like humans, and embodied machines. As corporations like Google invest heavily in technological and theoretical developments leading towards further, effective advances – a new ‘AI Summer’ – we can also see that hopes, and fears, about what AI and robotics will bring humanity are gaining pace, leading to new speculations and expectations, even amidst those who would position themselves as non-religious.

    Speculations include Transhumanist and Singularitarian teleological and eschatological schemes, assumptions about the theistic inclinations of thinking machines, the impact of the non-human on our conception of the uniqueness of human life and consciousness, representations in popular culture and science fiction, and the moral boundary work of secular technologists in relation to their construct, ‘religion’. Novel religious impulses in the face of advancing technology have been largely ignored by the institutions founded to consider the philosophical, ethical and societal meanings of AI and robotics.

    This symposium seeks to explore the realities and possibilities of this unprecedented apocalypse in human history.

    **

    You’ll note that thse two events address religious and ethical issues surrounding AI, which in turn revolve, I imagine, around the still disputed matter of the so-called hard problem in consciousness. I’d specifically welcome responses that explore any overlap between my title question and that hard problem.


    Switch to our mobile site