1 big thing: Forgetting to learn
In the quest to build AI that goes beyond today's single-purpose machines, scientists are developing new tools to help it remember the right things — and forget the rest, Kaveh reports.
Getting that balance right is the difference between a machine that can trade stocks like a pro but can't make heads or tails of a crossword puzzle, and one that learns all that plus a variety of other skills, and continually improves them — an important step toward human-like intelligence.
"AI is entirely about memory and forgetting," says Dileep George, founder of the AI company Vicarious.
- A computer that remembers too little won't be able to do anything that requires connecting past experiences to new ones — like understanding a pronoun in a sentence, even if the person it refers to was named just one sentence before. These memory lapses are known as "catastrophic forgetting."
- But one that remembers too much loses the ability to see the big picture. This is called overfitting: focusing entirely on the particulars of past experiences, at the cost of the ability to extract general concepts from them.
- "A big part of learning is knowing what to learn," says David Cox, director of the MIT–IBM Watson AI Lab. "You want to be able to forget things that are irrelevant." This holds for humans and machines both.
Another effect of catastrophic forgetting is that a computer learning a new task can lose the ability to do an old one — like a language learner forgetting their native tongue.
- To solve this, some researchers are adding memory modules that can set aside learned patterns, so that they don't get overwritten by new information.
- Others, like George, are experimenting with turning specific tasks into computer programs that are walled off from others and can be combined to perform more complex jobs.
These help with AI's forgetting problem — but they're not how human brains work, says Blake Richards, a neuroscientist and AI researcher at the University of Toronto.
- What our brains actually do isn't completely clear, Richards says. It's likely that memories are stored all together — but that patterns are kept separate from each other within the tangle, warding off the overwriting problem.
- Being able to connect memories with one another may be at the root of our ability to imagine and plan — two essential qualities that AI still lacks.
A trick humans do during sleep may be key to moving AI closer to the way we learn, says Cox. At rest, we relive recent memories, and in doing so reinforce neural pathways that help us remember them.
- Machines can mimic this with a process called "experience replay," which weaves in memories of previously learned tasks alongside new lessons.
- This helps them remember the old and the new — and because the memories are not kept separate, a computer could use parts of one to help learn the other, like a typist learning to play piano.
What's next: Perfecting memory could unlock AI "that can actually make insightful predictions and imagine what's going to happen in the future," Richards says. That's a crucial building block toward common sense, long a holy grail for AI researchers.
2. Six weeks without cash or cards
On April 7th, while vacationing in Florida, I lost my wallet, Erica writes.
At first, it really stung: I thought I would be totally incapacitated until I could get to the bank for a new debit card.
But it really hasn't been so bad. In fact, it has been so easy to live without cash or cards that I haven't made it to the bank in six weeks.
Why it matters: We've reported on how the U.S. has fallen behind on revolutionizing payments, keeping up its reliance on credit cards while China has leapfrogged from cash to mobile payments. Only 1% of Americans pay with their phones — that's fewer than the 2% who still use checks to pay.
But I found that a ton of the infrastructure to go cashless and cardless already exists. (Though it's worth noting that you do still need the card. You can just use it remotely.)
Here is what my six weeks turned up:
- My favorite breakfast spot takes Apple Pay.
- I can get any restaurant meal with UberEats.
- Or if I go out with friends, I can Venmo them.
- For groceries and toiletries, I can go to Trader Joe's or CVS (both take Apple Pay).
Sometimes, I really do need access to cash (to replace the Metro card that I lost, for instance). For things like that, I'll just go to Trader Joe's, buy a bottle of water, and request cash back at the register.
- Now that I've got the Metro card, I can just keep adding money to it online.
The bottom line: I actually kind of like using my phone to pay for things — it saves time and it's very, very convenient.
And cash is out. My former Axios colleague Martin Aguirre told me: "I paid for a coffee today in cash, and the woman looked at me like I was a drug dealer."
3. What you may have missed
Some distractions are worth it. To catch up, here is the top of this week's Future:
1. Beating the 'superforecasters': A geopolitical prognostication contest
2. The new sharecroppers: The hidden workforce behind the AI revolution
3. Untested systems for criminal justice: Much of applied AI doesn't work
4. The race to move stuff: Amazon wants to dominate another industry
4. Worthy of your time
5. 1 gorilla thing: New ape in Lincoln Park
Rollie, a western lowland gorilla, has given birth to a son, according to an announcement from Lincoln Park Zoo in Chicago. The birth last Sunday has attracted attention because poaching and the loss of their habitat have made western lowland gorillas critically endangered in central Africa.
- Here is a video of the yet-to-be-named newborn and his mom.
- “Even though this isn’t my first gorilla, I was really shocked and very excited," Jill Moyse, the zoo's curator of primates, told the Chicago Tribune.
This makes eight in the lowland gorilla troop at the zoo. There are three adult females, three juvenile females, plus a silverback male named Kwan.