“All known methods of algorithmic classification suffer from the problem of
overfit. Example: All training photos of a classroom have a blackboard, so
a room with a whiteboard can’t be a classroom. “
An intelligent computer algorithm runs in a well-managed virtual memory
space under the strict control of the programming. The algorithm is not
free to learn anything else. Human intelligence is programmed in many
ways, though it is ordinarily taught or trained by unwitting parents and
teachers.
If you want to transform data which can be logically and clearly defined
into some output which is equally well defined, algorithms will always win
hands down. If you want to transform potentially ambiguous, illogical and
unexpected data into somewhat desirable, adaptive but essentially
unpredictable and not entirely reliable output, you need a human as your
general purpose “machine”.
Our AI models are being constructed using the same outdated education
pattern. The neural networks are passive in their nature. They are “dumb”
and rigorous training has to be performed on them with huge data sets in
order for them to learn how to perform a certain task.
What we have seen lately, is that while systems can learn things they are
not explicitly told, this is mostly in virtue of having more data, not more
subtlety about the data. So, what seems to be AI, is really vast knowledge,
combined with a sophisticated UX.
What’s next? That’s easy. 2017 will be the year of artificial intelligence.
It is when every industry will ask how the latest old trend in computing
threatens to upend the existing order as they know it.
After a year of breakthroughs, experts believe they are on the brink of
revolutionising our daily lives through artificial intelligence – and Asia
can play a leading role in this brave new world.
The world’s biggest technology showcase kicks off in Las Vegas on 5 January
2017. Here’s everything we know about what will be happening at CES.
It is my suspicion though that this meta meta-model approach if pursued in
greater detail may the key in locking “Unsupervised learning” or
alternatively “Predictive learning” . Perhaps our puny human brains cannot
figure this out. However armed with meta-learning capabilities, it may be
possible for machines to continually self improve upon themselves. The
reason that this may not work however is that the vocabulary or language is
not available and furthermore somehow not derivable through this
bootstrapping method.
Anything can be automated. At least, that seems to be the industry
perspective. Humans are rash, impulsive and tainted by personal bias. So
why not just replace them with machines? Apparently, that’s what Facebook
thought when they fired their entire team of news editors for political
bias, replacing them instead with a new algorithm. That did not go well.
