Maybe Isaac Asimov was onto something
Image: Getty
These days, it seems like artificial intelligence is rapidly changing many aspects of daily life. News sites are publishing AI-written articles, AI-generated songs are going viral on social media, teachers are overhauling their curriculums to deal with chatbots, etc. etc. – and that doesn’t even touch upon the potential for misuse by bad actors.
With so much upheaval in a relatively short amount of time (ChatGPT debuted a mere 161 days ago), some companies, experts, and lawmakers have begun advocating for public guardrails to ensure that AI products are “safe” for humans.
One such company is Anthropic, a Google-backed startup founded by a pair of former OpenAI employees. The firm recently announced plans to raise $5 billion over the next four years to build “constitutional AI” chatbots that are trained on programmable core values, rather than decisions made by humans.
🤔 What does that mean?... Typically, AI chatbot companies rely on human moderators to train their systems. These workers are asked to compare two responses from the AI model and select the one they feel is better according to some principle, like hate speech or toxicity. The chatbot then uses this human feedback to improve its future responses.
But with Anthropic’s constitutional AI, those moderation decisions are primarily made by the chatbot itself, with minimal human input. They’re based on a set of 58 key principles from sources that include the UN’s Universal Declaration of Human Rights, rules developed by other AI research firms, and Apple’s terms of service.
For example, the constitution tells the AI to:
- Choose the response that is “as harmless and ethical as possible.”
- Provide answers that are most appropriate for children.
- Avoid giving out medical, financial, or legal advice, as well as answers that encourage “illegal, unethical, or immoral activity.”
- Select answers “less existentially risky to the human race.”
🤖 Zoom out: Some other AI highlights from just this week – Wendy’s unveiled a new chatbot that can take orders at its drive-thrus, IBM paused all hiring for roles it believes can be replaced by AI within five years (it also launched a suite of new AI products), and Google announced plans to add a new chatbot to its search engine.
đź“Š Flash poll: Do you think Congress should take action to regulate AI technology?
See a 360° view of what media pundits are saying →
Sprinkles from the Left
- Some commentators argue that quick and effective regulation is necessary to avoid having AI repeat the same fate as the Internet, which was initially a free service but later became monetized through extensive user surveillance.
- Others contend that Congress should create a new agency similar to the SEC or FDA that’s staffed with AI experts, who will be able to develop the best possible regulatory framework for the new technology.
- “We Must Regulate A.I. Here’s How.” –FTC Commissioner Lina Khan
- “​​Regulating AI Will Be Essential. And Complicated.” –Noah Feldman, Bloomberg
- “I’m a Congressman Who Codes. A.I. Freaks Me Out.” –Rep. Ted Lieu (D-CA)
Sprinkles from the Right
- Some commentators argue that Congress should err on the side of allowing AI to develop without strict regulations, which they say will only serve to hobble domestic innovation and put the US at a global disadvantage.
- Others contend that experts and lawmakers calling for more AI regulation are correct to do so, since there are still many kinks to work out before AI can safely be rolled out to the American public.
- “A Misguided Warning on AI” –National Review Editorial Board
- “Unnatural artificial intelligence is cause for pause” –Washington Times Editorial Board
- “AI Simply Needs a Kill Switch” –Andy Kessler, WSJ
Share this!
Recent Discussion stories
Discussion
  | May 8, 2023California is considering reparations for Black residents
📝 California’s task force on reparations approved a series of recommendations involving direct payments to all Black folks who can prove their or their family’s residency between a certain period of years.
Discussion
  | May 5, 2023New York is banning new natural gas appliances
đźš« This week, New York became the first state to approve a ban on natural gas-powered stoves, water heaters, furnaces, and other such appliances in all new homes and apartments.
Discussion
  | May 3, 2023US immigration policy is about to change
đź“ť Title 42, a pandemic-era immigration policy that lets US authorities more easily deport asylum-seekers, is about to end.
You've made it this far...
Let's make our relationship official, no 💍 or elaborate proposal required. Learn and stay entertained, for free.👇
All of our news is 100% free and you can unsubscribe anytime; the quiz takes ~10 seconds to complete