 |
| Image credit: Design Department Communicate Magazine |
The following is the editorial I wrote as an introduction to the Communicate magazine folder on AI which was covered in this issue (full disclosure: I am the editor in chief of Communicate). I thought it was fitting to republish it on the blog. Enjoy!
_______________________________________________
In the Steven Spielberg film “Minority Report” released in
2002, actor Tom Cruise portraying the character of John Anderton, is walking
into a mall, and is bombarded with ads that mention him by name, implying
they’ve been targeted specifically to him, and only him. Lexus and Guinness
beer to quote two of them. Perhaps because he is male and apparently of a
certain socio-economic class due to his dress, or perhaps they know he
researched those brands earlier. No matter, this was supposed to be “the
future”, an abstract, vague notion.
And whereas we are not there yet, perhaps we are not too far
off either.
The latest statistics dating back to October 2024 indicate
that there are 57,933 Artificial Intelligence (AI) startups in the world. The
global AI market is valued at $196 Billion globally, and the AI industry is
projected to grow by 13 times till 2030. Neflix, the streaming company makes $1
Billion annually from AI-based recommendations by letting AI hone algorithms
keeping viewers coming back for more. Interestingly, 48% of businesses use
machine learning (ML), data analysis, and AI tools. AI software ChatGPT reached
1 million users in 5 days – and that was only 2022. By 2023 they had 100
million monthly users.
87% of global organizations believe AI will give them a
competitive edge, and during the period between 2024 and 2030, AI is expected
to achieve a compound growth annual rate of 37,3%. In 2020, AI generated
services generated $19,6 Billion in 2025 this number is expected grow to $126
Billion. Within that same year, 97 million people are expected to be working
within the realm of AI. On the ground though, automated emails and chatbots are
two of the most entry-level ways companies use AI. Interestingly, AI is
forecasted to increase employee productivity by 40%. Apparently 49.21% of
marketers believe AI for email market revenue generates more growth.
Not a day passes without a startup announcing novelty
technology related to AI. Take Osmo, the brainchild of former Google researcher
Alex Wiltschko which focuses on detecting smell. Sure, smell has been a main
component of detecting diseases in the medical field, but practically, Osmo
also partnered with an unnamed yet well-known second-hand sneaker reseller to
authenticate its goods and it had a 95% success rate.
However, on the other side of the coin employees in certain
sectors have fears of being made redundant, and such fears might be justified
for workers in the transportation and storage, manufacturing, and wholesale/retail
industries. The list of sectors might get longer as time goes by.
When the industrial revolution happened, the premise was:
Let the machine do the work, the business will expand, and all the people who
lost their work will be brought back to the pool. Obviously, that never
happened, in the case of AI, apparently there is a loss of 7% of jobs but an
addition of 9% - though not the same people for sure.
Voice assistants, which basically use Natural Language
Processors – or the ability to translate human speech into information
computers can understand – are one of the most used forms of AI. Words like
Siri, or Alexa have entered the lexicon. Even smartwatches that indicate
specific times to do certain activities (such as washing hands during the COVID
pandemic) are also built on the AI principle.
But before we put the horse in front of the carriage and get
carried away, let us contemplate this example which Apple released very
recently:
“Oliver picks 44 kiwis on Friday. Then he picks 58 kiwis on
Saturday. On Sunday, he picks double the number of kiwis he did on Friday, but
five of them were a bit smaller than average. How many kiwis does Oliver have?”
Obviously, any person familiar with basic maths will say –
44+58+(2x44) = 190. AI models apparently, went with 185. Why? Because they were
tricked by the “five of them were a bit smaller than average”. Of course, a
human – even a child - knows that a kiwi “smaller than average” still counts as
one kiwi, an AI model got confused by the detail.
The bottom line is that AI models are good at mimicking data
they were previously fed, but when it comes to “abstract reasoning”, AI models
fall short. To put it bluntly, at this point, they are not “intelligent”.
Perhaps we are asking too much of a nascent technology.
Maybe with time, machines will get to understand that five kiwis “smaller than
average” are actually, still kiwis. Just as right now the AI industry is growing
by leaps and bounds with advertisers, producers, film makers, and whatever
other specialty in the communication field are trying to harness its power and
produce whatever is producible within the limits and hopes of the current AI
technologies.
There was a full ad for a new car produced in one day using
nothing but AI technology. Was it boring? Yes, it was just beauty shot after
beauty shot with little in what can be considered as “idea”. Surreal films
abound everywhere on the net to show the prouesse of what AI models can do –
from bears in trams to creatures morphing into other shapes, yet none of them can
hold a candle to A Chien Andalou in terms of emotional punch. For reference Un
Chien Andalou is a 1929 French silent short film directed, produced and edited
by Luis Buñuel, who also co-wrote the screenplay with Salvador Dalí. Perhaps
the difference lies not just in getting the execution “right” but in the
creative backbone behind the works in question.
Or it could be that the AI models still need a human
“prompt” behind them. To initiate commands and mimic actions, and “play
pretend”. Once more, no one can truly grasp how big this can go. To go back to
a cinematic analogy, in the film Blade Runner released in 1982, the Los Angeles
Police Department ran the “Voight-Kampff” test to detect Nexus-6 replicants. Deckard,
the main character and who identified the Nexus-6 (and played by Harrison Ford)
was according to director Ridley Scott… A replicant himself. An AI identifying
AI?
Which brings us back to the origin of all of this – the Alan
Turing test. One that led him to believe that "machines can think". We
could be very close or very far from this scenario.
Right now, to differentiate ourselves from robots, we need
to identify images with traffic lights in them.