Stocks at first traded cautiously as Federal Reserve Chair Janet Yellen testified before Congress, leaving an opening to a March rate hike. The major indexes ended solidly higher at new record highs, with the S&P 500 Index rally fueled particularly by financials.
The Dow rose 92 points, with 22 of its 30 components gaining ground; the S&P 500 Index added 9; and the Nasdaq closed higher by 18. Advancers narrowly led decliners on the NYSE and led four to three on the Nasdaq. The prices of Treasuries weakened. Gold futures fell 40 cents to close at $1,225.40 an ounce, and the price of crude oil rose 27 cents to settle at $53.20 a barrel.
In other business news:
- Bond yields rose and stocks were at first wobbly as Federal Reserve Chair Janet Yellen gave regularly scheduled testimony to Congress on the Fed’s monetary policy. Yellen’s remarks left open the possibility that the Fed would raise interest rates at its meeting in March, a more hawkish position than many analysts had expected. Yellen emphasized the risks of waiting too long to raise interest rates in the face of a growing economy, saying it would be “unwise” to try to play catch up.
- U.S. producer prices jumped more than expected in January, according to the Labor Department’s Producer Price Index. The index gained 0.6% for the month, the biggest jump in four years. Much of the monthly gain came from surging gasoline prices, up 12.9% for the month. Excluding energy and food, however, the Core Index still gained a robust 0.4% for the month. Year over year, producer prices were up 1.6% overall and 1.2% for the core.
- Following a judge’s antitrust ruling, Aetna Inc. and Humana Inc. officially called off their proposed $34 billion merger, deciding not to appeal the ruling. Humana said it will receive a $1 billion breakup fee. The judge in the case found that a merger would particularly hurt seniors by reducing competition in the Medicare Advantage market. Aetna’s shares (AET) rose 3.10%, and Humana’s (HUM) fell 0.38%.
- Small business confidence moved slightly higher in January to the highest point since December 2004, according to the National Federation of Independent Business. The organization’s small business confidence index rose to 105.9, slightly higher from December’s reading of 105.8. Analysts had expected a slight pullback from December, when the index recorded the largest monthly increase in its history, jumping 7.4 points from November.
Today in AI:
–Artificial Intelligence promises to be as revolutionary as the nineteenth century’s Industrial Revolution, which is saying a lot because the latter has “revolution” right in its name. And just like the Industrial Revolution, the advent of AI is proceeding without (much) concern over installing rudimentary safety equipment.
Case in point: Google is working on an artificial intelligence program called DeepMind, made famous last year when it beat a human player at the game of Go. Now that DeepMind has a few wins under its belt, Google is testing it to see not just how it wins, but how far it will go not to lose. And as Google discovered, it’s willing to get very aggressive.
In a new experiment, Google pitted different versions of DeepMind against each other in a simple computer game that involves gathering digital apples. As with most games involving gathering apples, this one also features lasers. As long as there were plenty of apples, the DeepMinds cooperated with other. When the apples ran low, they brought out the lasers to temporarily knock each other out so they could gather more apples than their opponent.
Here’s the scary thing: Google ran the test with different versions of DeepMind, some simple, some more complex. When the simple versions faced each other, they were more likely to be happy with a draw, with each DeepMind gathering the same amount of apples. The more complex versions, however, ones with deeper and more involved neural networks, whipped out the lasers as soon as the apple supply diminished. In other words, AI with a goal can be very aggressive in pursuit of that goal, and the smarter it is, the more aggressive it might be. Maybe instead of creating super-smart AI that can beat the best Go players in the world, we should settle for an AI that’s content to dominate Chutes and Ladders (and whatever we do, let’s keep it away from Risk).
–At least we’ll always be able to out-bluff computers in poker, is something we would have been able to say before last month. In January, an AI poker-playing program from Carnegie Melon beat four online poker professionals in Heads-up No-Limit Texas Hold ‘Em over the course of 120,000 hands played over 20 days, ending with a convincing $1.7 million margin of victory. This isn’t like counting cards at Blackjack or cranking out calculations about the finite (though large) number of possible moves in a chess match. The AI had to bluff. It also had to regularly update its strategy so its human opponents couldn’t figure out its tells over the marathon tournament. The days of shoddy online poker programs might be over. (“Look, its hourglass icon has been spinning for four seconds. That means it doesn’t even have a pair.”)
According to Professor Tuomas Sandholm, one of the designers of the AI, “The best AI’s ability to do strategic reasoning with imperfect information has now surpassed that of the best humans.” Great. We can look forward to a future of robots with an aggressive streak who can out-bluff humans. Whatever we do, let’s not give them lasers.