News

Most data is ‘crap’ so don’t bother with data scientists, says former US intelligence officer

Marketers should not waste their time or money hiring data scientists as the information businesses are collating is often not good enough to justify their high salaries, a former US intelligence officer and data analyst has said.

drew-perez-finance-marketing-2016

Drew Perez: “This is not Terminator Judgement Day”

 

Drew Perez, founder of tech firm Adatos and a 26-year veteran of the US Department of Defense, also warned marketers not to view machines, and more specifically artificial intelligence, with trepidation and stressed machines are no substitute for humans when it comes to creativity.

“This is not Terminator Judgement Day,” he told delegates at the Mumbrella Finance Marketing Summit. “It’s up to us humans what we let machines do.”

Perez, in an address titled Rise of the Machines, said too many businesses have “dirty” data, telling the conference that a data scientist “is the last person you need to hire”.

“Do not hire a data scientist despite what the Harvard Business Review says,” he said. “It is said to be the sexiest job in the 21st Century. If you want to be sexy, go for it.

“But the reason [for not hiring a data scientist] is that they don’t scale. Second, if the data is crap, dirty, not indicative – and just because you have a lot of data does not mean that it’s indicative or relevant – they don’t have a job.”

He said data scientists are paid salaries of US$200,000 or US$500,000 “but they can’t do anything” but clean the data and get it ready to “train the machine”.

“Artificial intelligence is basically this. You train a machine to think,” Perez said. “To do that you had better be a good parent and send your child to the best school.

“Send them to a prep school and give them a crap education is going to give you crap, and that is the elephant in the room.

“Because you hire a data scientist does not mean you are going to get anything but crap if your data is crap.”

Asked later to clarify his comments, Perez said scientists were not a waste of money, but repeated his assertion that they “don’t have work to do if your data is not indicative”.

When companies receive data there are three scenarios, he said.

Perez: "There's a morality to machines as well"

Perez: “There’s a morality to machines as well”

“Is the data rich enough to be indicative, and if it’s not you’ve got nothing. Maybe it is, but you may have to beg, borrow and, if you’re a government, steal to enrich it. The third outcome is yes, you do have something and then you can do something with a data scientist.

“But for the most part…it is highly probably that you don’t have indicative data.”

Perez warned marketers on two fronts: not to distrust machines or think they will solve creative problems.

“If you don’t trust the machines you are not in the game. If you think it’s all Terminator Judgement Day you are missing the point,” he said. “But don’t misunderstand it or have the wrong expectation because if you want AI to be creative in an artistic way, this is not what machines are good for.

Use machines for what they are good for, and that is menial and repetitive tasks at extremely low cost. Use humans for what we are good for. It’s a partnership. If you know how to blend it to your advantage you have a winning team.”

Perez also stressed that companies have a responsibility not to abuse the power of machines at the expense of the workforce.

“Whenever I train someone in my company the last thing I go through is the morality,” he said. “Yes I can deploy a box to replace a call centre with 10,000 trunk calls but is it responsible for me to deploy technology like that that displaces 10,000 jobs in a developing nation?

“You play this game, you have responsibilities.”

ADVERTISEMENT

Get the latest media and marketing industry news (and views) direct to your inbox.

Sign up to the free Mumbrella newsletter now.

 

SUBSCRIBE

Sign up to our free daily update to get the latest in media and marketing.