Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/why-ai-is-pseudo-science-hype-shown-by-cancer-detecting-chip-pr.8246/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/JobRunner] => 2000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000670
            [ThemeHouse/XPress] => 1010394
            [XF] => 2020270
            [XFI] => 1040070
        )

    [wordpress] => /var/www/html
)

Why AI is pseudo science hype - shown by Cancer detecting chip PR

There is interesting PR from IBM research press release on new chip
to detect cancer. Claim is that chips using bio-assay can detect
cancer. Here is a URL for one story:

IBM Lab on Chip Detects Cancer | EE Times

Area is interesting because there is a very complex bio-marker
test for diagnosing Colon cancer that is called Cologuard marketed
by a company called Exact Science. It went through decades of research
by the best medical researchers (mostly from Mayo Clinic) and has
FDA approval and is approved for Medicare re-reimbursement but even then
Medical establishment is undecided on whether it is better than
colonoscopies. Test involves sending preserved feces sample to
company laboratory.

AI and robotics has this same problem. It is science (maybe
engineering) by press release. For example how does deep learning
compare to traditional search methods? Where are the double blind
experiments?

Another example is the claim that computers can beat the world's
best chess players. Consider the reward/risk situation for a chess
player. Lose and you make millions. Win and you go back to playing
low prize money chess tournaments. I think the following double blind
competition would have opposite result. Chess player does not who
opponent is. If player loses to a low ranked player, it is counted
to minus players chess rating.

I have run into another situation from my claimed solution to the P=NP
problem showing that P is equal to NP (ref. [1603.06018] Philosophical Solution to P=?NP: P is Equal to NP).
I show that Turing Machines are the wrong model for computation. The right
model is called MRAM (random access machine with unit multiply) that
is the model Von Neumann used in developing the first digital computers.

It shows that that there can not be better computer programs than
traditional Von Neumann computing. All the AI research assumes the
Turing machine model.

No one is looking for better algorithms than PR department named
"Deep Learning". There are 113 claimed solutions to the P=?NP problem
that all involve interesting algorithms, but never make it past blind
reviewers to publication. Maybe because the $1 million Clay Institute
Prize requires refereed publication. The solutions are collected here:

P-versus-NP page
 
Couldn't agree more if read as "a lot of publication on AI is pseudo-science hype". A direct result (in my view) of the click-driven deluge of information, which has us all exaggerating the importance of whatever we have to say, simply to try to rise above the noise. Of course AI has had some localized successes, particularly in vision and in niche game-competition (chess, Go) which, when pushed through the media significance amplifier, has us all being replaced by machines in the near future. I'm all for blue-sky thinking but we have to spend most of our time closer to the ground if this stuff is ever going to have widespread impact.
 
Last edited:
Top