Ai is not what you think

This article is part of the On Tech newsletter. You can do this Register here Getting it on weekdays.

When you hear about artificial intelligence, stop imagining computers that we can, but can do better.

My colleague Cad metz, Who has New book Regarding AI, we want to understand that technology is promising, but it has its downsides: it is currently less capable than people, and is being coded with human bias.

I talked to Cade about what artificial intelligence is (and is not), areas where he is hopeful and frightened of the consequences and areas where AI is less than optimists’ hopes.

Shira: Let’s start with the basics: What is artificial intelligence?

Cad: It is a term for a collection of concepts that allow computer systems to function vaguely like the brain. Some of my reporting and my book focus on one of those concepts: a neural network, which is a mathematical system that can analyze data and pinpoint patterns.

If you take pictures of thousands of cats and feed them into a neural network, for example, it can learn to recognize patterns that define what a cat looks like. The first neural network was built in the 1950s, but they never fulfilled their promise for decades. This began to change around 2010.

what changed?

For decades, neural networks had two important limitations: neither sufficient data nor sufficient computer processing power. The Internet gave us data, and ultimately scientists had enough computing power to do all of this.

Where can people see the impact of neural networks?

This single idea changed many technologies over the last 10 years. Alexa, digital assistants such as driverless cars, Chat bot, Computer Systems that can write poems, monitoring system And Robots that can lift products in warehouses All rely on neural networks.

Sometimes it seems that people talk about artificial intelligence as if it is a magic potion.

Yes. The original sin of AI pioneers was that they called it artificial intelligence. When we hear the word, we imagine a computer that can do anything that people can do. This was not so in the 1950s, and this is no longer true.

People do not realize how difficult it is to imitate human reasoning and our ability to deal with uncertainty. A self-driving car can recognize what’s around it – in some ways people can do better. But it does not work well to drive anywhere at any time or what you and I do, like some amazing reactions on the road.

What are the downsides of neural networks and AI?

Many. Machines will be able to generate massive misinformation. There will be no way to tell what is online and what is fake. Autonomous weapon It also has the potential to be incredibly dangerous.

And the biggest thing is that many companies have promoted the algorithm as a utopia that removes all human faults. It is not. Some neural networks learn from enormous amounts of information on the Internet – and this information was created by people. That means we are Creating computer systems that demonstrate human bias For example, against women and people of color.

Some American technologists, including former US CEO Eric Schmidt Say The United States is not taking AI seriously, and we risk falling behind China. How real is this concern?

It is valid but complex. Schmidt and others want to try to ensure that the most important AI technology is built inside the Pentagon, not just inside Google’s giant technology companies.

But we have to be careful about how we compete with a country like China. In the United States, our best technology talent often comes from overseas, including China. Closing our borders Experts in this field can harm us in the long run.


Tip of the week

A reader named Eva emailed him asking for small software programs known as browser extensions, plug-ins, or add-ons for Chrome, Safari, and Firefox that he claimed would save him money.

“I saw advertisements for these browser add-ons like Honey (from PayPal) and Capital One Shopping,” she wrote. “They claim that whenever you shop online, they will automatically find and apply a promo code to protect you. It sounds great, but I keep wondering what is in it for them? They just do it for themselves Not doing it for the good of the heart. Before I sign up for these services, I want to know what the trade-off is. Can you help me?

Brian x. Chain, New York Times personal technology columnist, has this response:

Yes, there is always a trade off. With free software, your personal data is often part of the transaction.

I should take a few minutes to research the company’s business model and privacy policy.

More than a year ago, Amazon warns customers to remove honey add-ons Due to privacy concerns. Honey’s Privacy Policy States: “Honey does not track your search engine history, email, or your browsing on any site that is a retail website (not a site where you can shop and shop).”

Read between the lines: This means that Honey can track your browsing on retail websites. (Is honey said It only uses data in ways that people expect.)

Privacy Policy for Capital One Shopping More clearly: “If you download and use our browser extension, we may collect browsing, product and e-commerce information, including product page views, pricing information, location data, various merchant websites and services. But not limited to purchase history, the price you paid for the items, whether a purchase was made, and the coupons you used. “

This is a lot of information to hand out for software that automatically applies coupons. Fair trade is up to you or not.


  • Hence. very. Money. everywhere: My colleague Erin Griffith Connects the dots Between the sale of digital art for $ 69 million, a frenzy for cryptocurrencies and rising prices of things like vintage sneakers. Basically, it pays to take financial risk right now, with our minds turning to Goo in an epidemic. related: Stripe, which makes software plumbing for businesses to accept digital payments Now one of the most valuable start-ups in history.

  • Facebook is studying our commentary ideas: Facebook is conducting internal research on the spread of ideas on its app that contribute to vaccination, The Washington Post Reported. Initial findings suggest that messages that are not false at all can “cause harm in some communities where it has a resonant effect,” Post said.

  • How to Protect Americans: Failures of US intelligence agencies to detect recent digital attacks by Russia and China are causing US officials Rethink how the nation should protect itself, My colleagues reported. One thorny idea is for tech companies and US intelligence agencies to collaborate on real-time assessments of Cyberheart.

Go Hug a cow. This can help.


we want to hear from you. Tell us what you think about this newspaper and what you want us to find out. You can reach us ontech@nytimes.com

If you don’t already find this newsletter in your inbox, Please sign up here.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *