- Hackers successfully evaded programs using LeetSpeak.
- Captcha was used to prevent bots from posting auto-messages and prevent accessing personal information on emails.
- Google’s self driving car- WAYMO
Humans are training the AI Supercomputers, says an article published in Medium.
How data was protected in early internet days?
As people stored data digitally during early internet days, there was an inherent caveat in storing such data. They are:
- Identity theft
- Data theft
- Altering Data and misleading users.
If sensitive information was posted online, the computers would interpret the keywords and block them out (or) delete the content. The entire internet industry has been battling against those who alter these data, called Hackers. Anti-virus software, firewalls, etc., have been introduced to protect systems & data from these hackers.
How hackers bypassed the computer interpretation?
In the early days, hackers made texts illegible so that computers could not interpret them.
For example: to bypass the computer interpretation, the word
HELLO would be written as
|-| 3 |_ |_ 0
# 3 1 1 O
By this, the hackers successfully evaded programs that could block them out. These are called ‘LeetSpeak,’ where the alphabets are modified either as numbers or symbols to bypass the filter program detections.
Each of the alphabets can be ciphered as numbers, symbols, or a combination of all, such that computer filters cannot interpret or block the hackers out.
Some examples/tables of LeetSpeak
CAPTCHA – Completely Automated Public Turing Test to tell Computers and Humans Apart
To prevent bots from posting auto-messages or comments and prevent accessing personal information on emails, CAPTCHA came into existence.
A hacker bot cannot interpret the Captcha, and only humans can decipher. Thus Captcha was initially used to differentiate between bots and humans. Companies like PayPal used these as a part of their fraud detection strategy.
During 2005 ~ 2010, Google acquired Captcha and came out with reCaptcha. Google achieved its dual purpose of personal identity or data protection and digitizing google books and other online magazine archives.
For example, recognizing words was first given to the AI supercomputers, and those answers were recorded. The same set of words was displayed in the reCaptcha so that humans could enter their response, which again is cross-checked with the supercomputer’s answer.
The prime reason is to separate bots and humans, yet, another reason is to train the AI Supercomputers for image and object recognition.
reCAPTCHA also makes positive use of the human effort spent in solving CAPTCHAs by using the solutions to digitize text, annotate images, and build machine-learning datasets. This in turn helps preserve books, improve maps, and solve hard AI problems.
From the above image, we check the squares where the fire hydrants exist, thereby training the AI supercomputers — cross-checking with billions of users. So, we are doing these challenges by providing labeled data used in a training dataset for some AI under the Alphabet Inc (Google) umbrella.
The image doesn’t need to have something to check — at times, as seen in the above photo, if the AI did not detect any signs, it would be presented to humans to get it verified, to ensure:
- What humans can see, Can the AI/Computer vision can see as well?
- What AI/Computer vision cannot see, can the humans see it? Or Vice-versa!
In 2014, Google pitted one of its machine learning algorithms against humans in solving the most distorted text CAPTCHAs: the computer got the test right 99.8 percent of the time, while the humans got a mere 33 percent.
How does Google verify the selected images?
When Google presents you with a panel of, say, six images, five of the images are already labeled. The web user is asked to identify five images correctly, including Google is looking to label. We only need to identify the four images Google already has labeled correctly, and our answer for the fifth unknown image goes into the AI training dataset.
We train the AI supercomputers to recognize objects on the road and streets, only to help Google maps show them on the apps.
Words / Numbers → Captcha → to separate humans and bots
Words from Books → reCaptcha → to digitize books → Google Books
Words → reCaptcha → Google Translate
Objects / Properties → Google Lens
Street names / Door signs / building names → Google maps / Street view / Live View
All these are leading to next-generation technology → WAYMO → the self-driving vehicles.
Did you subscribe to our daily Newsletter?
It’s Free! Click here to Subscribe