AI as Super Spam Filterer

Table of Contents

I was once asked for my favorite example of AI in daily life, one that consumers take for granted or don’t realize is made possible by AI. My answer: spam filters.

We’re all familiar with spam. Named after the meat globby trademarked product, spam are unsolicited email messages, the electronic equivalent of junk mail and like what meat lovers think about the stuff that comes in that metal tin of SPAM®. (Do you know the name for something that’s not spam? Ham. But I digress.)

Spam email first made its appearance in 1978 when marketing manager Gary Thuerk sent out unsolicited bulk email to promote a new model of computer. Today, estimates of the percentage of emails that are spam range but they’re all large numbers — as much as 85 per cent of the 333.2 billion emails delivered everyday. Of that spam email, one estimate at the end of 2021 was that almost 25 per cent were generated in Russia.

Fortunately, machine learning spam detectors allow email providers to filter out all that junk for us, their users. While imperfect, like all things AI, the detectors have been improving and will keep getting better.

Machine learning algorithms trained by AI systems use statistical models to classify data. When it comes to detecting spam, which can range from annoying unsolicited junk to harmful phishing emails, fraud scams and malware attacks, these machines analyze words and word combinations in an email to determine which are spam and which ones are ham.

For instance, AI can detect “spammy” words, symbols and numerals in an email such as “free,” “#1” and “cash bonus.” Spam filters look for patterns in the content of messages.

Of course, nothing’s perfekt. These systems give false positives and need to be re-trained to perfect their ability to detect spam. For instance, it’s estimated that about three quarters of email messages are known as graymail — email that is identified as spam may be legitimate, such as newsletters, social network updates and company offers. AI systems need people to ensure these things don’t slip through the cracks.

And as with most things in life, the bad guys are always trying to stay one step ahead of the good guys behind AI. For instance, scammers will embed special characters or spacing in words l i k e  t h e s e or embed HTML codes to try to trick the spam detectors. That’s where email providers rely on users to continually improve and update their spam detection systems. Human and machine complement one another.

If you find that spam emails are still getting through into your inbox, know that the vast majority of spam emails never make it there. Spam filters are always evolving to keep up and keep out those billions of unwanted emails as are the filtering rules we apply to suspect words and messages.

Now, what’s for dinner? I smell ham.

Subscribe to our newsletter

Keep up-to-date with the latest developments in artificial intelligence and the metaverse with our weekly newsletter. Subscribe now to stay informed on the cutting-edge technologies and trends shaping the future of our digital world.

Neil Sahota
Neil Sahota (萨冠军) is an IBM Master Inventor, United Nations (UN) Artificial Intelligence (AI) Advisor, author of the best-seller Own the AI Revolution and sought-after speaker. With 20+ years of business experience, Neil works to inspire clients and business partners to foster innovation and develop next generation products/solutions powered by AI.