AI For Everyone: What Andrew Ng wants to convey with this Non Technical Course in 30 points.

AI For Everyone: What Andrew Ng wants to convey with this Non Technical Course in 30 points.

Harveen SinghBlockedUnblockFollowFollowingFeb 28SourceAI for everyone is a non technical course taking which you will have greater knowledge than most CEO’s in the world.

At least this is what Andrew Ng claims.

So let’s find out in short what he wants to convey.

AI to create 13 Trillion Value by 2030 mostly to be used in Retail followed by Travel and Automotive sector.

AI is broadly categorized into ANI (Artificial Narrow Intelligence) and AGI(Artificial General Intelligence).

With a lot of progress in ANI people falsely started believing that they are progressing in AGI.

Don’t spend much on the IT infrastructure to collect data.

Feed data as early as possible to AI team so that they can figure out whether that collected data is useful and can change the Data collection strategy.

Also it is not so that more the data, more the value!Machine Learning is all about learning A to B mapping where A is the input and B is the output label whereas Data Science is more about extracting insights and conclusions from data .

The output in case of Machine Learning is software whereas in case of Data Science is a slide deck.

Deep Learning is a branding name for ‘Neural Networks’ that are nothing but big Mathematical equations.

Neural Networks were inspired by brain but the internal functioning is almost unrelated to how actual brain works.

Just as: Shopping mall + Internet != Internet companySimilarly:Any company + Deep learning != AI company.

Any problem what a human can do with 1 second of thought and for which lots of labelled data is available can be automated with supervised ML.

For example- if a user will click on add or not.

AI cannot empathize or understand gestures at the moment.

AI cannot learn complex task with small amounts of data.

For Machine Learning:Collect Data, Train Model and Deploy model.

 For Data Science:Collect Data, Analyze Data, Suggest Changes.

For example: In recruitment, Data Science will help us to optimize the recruiting process by analyzing data.

Whereas machine Learning can help in automated resume screening.

Select Projects that are feasible and both valuable for your business.

While deciding a project both AI experts and Domain Experts should work together.

Automate tasks not jobs.

Understand Pain Points in your business.

You can make progress even without big data.

In addition to Business Diligence and Technical Diligence, think of Ethical diligence as well whether the project you are building will bring some good to Humans.

To the AI team, specify your statistical acceptance criteria on the test set.

Roles: Software Engineer: write software code like a function/subroutine.

Machine Learning Engineer: Responsible for creating modelsMachine Learning Scientist: Responsible for extending state of the artApplied ML Scientist: A role in between ML Engineer and ResearcherData Scientist: Examine Data and provide insights to drive business decisionsData Engineer: Make sure data is easily accessible in a secure and cost effective wayAI Product Manager: What to build, whats valuable and feasibleSourceExecuting Relevant Pilot AI projects can set the traction for 6–12 months.

Create one central AI team and disperse it to multiple Business Units under the leadership of CAIO (Chief AI Officer).

Initially the CEO should provide the funding to AI unit rather than a BU providing the funding and after the initial investment AI team has to show its value that is creating for the BU.

Business Leaders must understand what AI can do for their enterprise.

AI Team leads should set project direction and monitor resources.

In house AI engineers should be trained to work on AI pipeline.

CLO should know how to curate content rather than create content.

Build an AI strategy only after executing one or two projects or it will come up as an academic strategy not practical strategy.

Different companies have different strategies.

A good product started with less data will have users.

Over the time these users will generate data which can be used to improve the product and so on.

Strategic Data Acquisition.

Don’t monetize product for collecting useful data.

New roles like Machine Learning Engineer should be promoted.

Pair Engineering Talent with Business/Sales Talent to find feasible and valuable projects.

Don’t expect AI project to work the first time and don’t enforce traditional planning processes in an AI project.

Get friends to learn AI, brainstorm projects and find a mentor!Neither be too optimistic about AI that super intelligence is coming.

Neither be too pessimistic about AI that AI winter is coming !.Be somewhere in middle!Explainability of AI is hard.

AI can become biased with biased data.

AI systems are open to Adversarial Attacks.

In future companies might be at war with the adversarial attackers.

US and China are leading in AI but this technology is still immature giving other nations an equal advantage to compete.

By 2030 according to report by McKinsey & CompanyJobs displaced by AI : 400–800 MillionJobs created by AI : 555–890 MillionThank you Andrew Ng!.Overall I liked the course, I wish there could have been more for Human Resources professionals who should understand tools like tensorflow, keras etc.

But once again, It was good to see Andrew Ng back in action.

Just a last joke to finish it off!Why are there so many shocking results in AI?Because AI is the new electricity,Shocking electricity!! :DRead my other Articles On Medium:My first Data Science InterviewMachine Learning: What, When and How?Kalman Filter InterviewAbout Me: I work as a Self Driving Car Engineer focused on providing Intelligence to Vehicles using Deep Learning.

Please reach out to me on LinkedIn, its good to connect with people.


. More details

Leave a Reply