AI and Machine Learning – Notes from the wild side

The impact of AI on the market today is all about the solution.  But tomorrow – let’s say 2020, 85% of all online actions will be done through a machine.

AI is the study of computers performing tasks that could otherwise have been done by humans.  The main areas being looked at are:

  • Reasoning
  • Knowledge representation
  • Planning and Predictive Analytics
  • Machine Learning – learn from consequences
  • Communication – natural language processing, speech to text etc
  • Perception – visual recognition
  • Motion and manipulation – robotics

We are currently in ‘Weak AI’ mode in that it has to be subject matter specific.

Where it is going is ‘Strong AI’ mode in that you have full human intelligence, learning to reason and communicate etc.

People seem to think AI is all about the tech.  Well I am telling you it isn’t.  You need to engage behavioural psychologists, developers, analysts, math modellers amongst others.

Types of data which can be used

  • structured data
  • unstructured data

Text analysis

  • Question answering
  • Personality insights
  • Sentiment analysis

Image Analysis

  • pattern analysis for skin cancer – one such project has 41,000 images of moles and the image analysis is being used to detect rogue moles
  • Oil pipeline checking – run a robot over lines continually to see weakness and learn from the degradation in images whether it is due for renewal or the probability of breakage

Numerical Analysis

  • Trending probabilities
  • Anomaly analysis – would be great to use in conjunction with cyber security breaches to look for behaviour that is different from the norm

Predictive Analysis

  • Risk analysis and predictive reasoning, similar to the human Actuary
  • Sales funnel plug-ins giving sales reps the lead for customers who will most likely buy – I believe SalesForce has a plugin for it

Digital Concierge

  • Consistent messaging
  • Direct and instant responses
  • Faster resolution

Machine Learning

  • Excellent for digital marketing to take human bias away
    • How much
    • How many
    • Which category
    • Which group
    • Adaptive algorithms


  • Facial recognition – reception AI robot recognising your face and searching for your meeting then notifying the organiser you have arrived
  • linking to outlook calendar
  • Delivers brand experience
  • Complements not replaces resources
  • Repetitive tasks
  • Dangerous tasks


Oh, and one important fact – when you release an AI or machine learning programme into the wild – it is like a child, it is never going to be good from day one – it needs to learn to get better, it needs continual training and learning.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s