I’m not sure that anything so inconsistent truly passes the Turing test. Even Donald Trump would be questionable.
ChatGPT has passed the Turing test and if you're freaked out, you're not alone
Despite just releasing ChatGPT-4, OpenAI is already working on the fifth iteration of the immensely popular chat application, GPT-5. According to a new report from BGR, we could be seeing those major upgrades as soon as the end of the year.
One milestone, in particular, could be within reach if this turns out to be true: the ability to be indistinguishable from humans in conversation. And it doesn’t help that we’ve essentially been training this AI chatbot with hundreds of thousands, if not millions, of conversations a day.
Computer and AI pioneer Alan Turing famously proposed a test for artificial intelligence that if you could speak to a computer and not know that you weren't speaking to a human, the computer could be said to be artificially intelligent. With OpenAI's ChatGPT, we've certainly crossed that threshold to a large degree (it can still be occassionally wonky, but so can humans), but for everyday use, ChatGPT passes this test.
(Related) Another perspective.
https://www.exponentialview.co/p/chartpack-measuring-ai-1
Chartpack: Measuring AI (1/3)
The mismeasure of AI: How it began
What if the way we evaluate artificial intelligence was flawed?1 The rapid rise of ChatGPT and other large language models (LLMs) has left us struggling to understand where we stand in the AI landscape. Old standards, like the problematic Turing Test2, are no longer relevant, with GPT-4's output already being (mostly) indistinguishable from human-made text. However, this doesn't mean that it has reached human-level intelligence, only that it can mimic our outputs. Even OpenAI’s Sam Altman deemed it "a bad test" for these models.
This leaves us in a predicament. How do we understand the capabilities and impacts of these models?
I see this as a plea for time to catch up.
https://cointelegraph.com/news/elon-musk-and-tech-execs-call-for-pause-on-ai-development
Elon Musk and tech execs call for pause on AI development
… The institute called on all AI companies to “immediately pause” training AI systems that are more powerful than GPT-4 for at least six months, sharing concerns that “human-competitive intelligence can pose profound risks to society and humanity,” among other things.
Does this resolve anything? Is $725,000,000 even material?
https://www.pogowasright.org/judge-approves-725-million-deal-in-meta-data-privacy-class-action/
Judge approves $725 million deal in Meta data privacy class action
Natalie Hanson reports:
A federal judge on Wednesday granted preliminary approval of Facebook parent company Meta’s agreement to pay $725 million to a class of millions of people whose personal information was harvested in the Cambridge Analytica scandal.
Meta Platforms agreed this past December to pay $725 million to settle claims by its users that the social-media behemoth illegally gave third parties, including political consulting firm Cambridge Analytica, access to their private information.
The settlement marks the largest recovery ever achieved in a data-privacy class action, and it is the most Facebook has ever paid to resolve a private class action.
Read more at Courthouse News
I may need the Privacy Foundation’s April 21st seminar to make sense of these laws. I am unable to do it on my own.
https://www.pogowasright.org/idaho-is-about-to-become-even-more-extreme-on-abortion/
Idaho Is About to Become Even MORE Extreme on Abortion
Alanna Vagianos reports:
Idaho already has some of the most extreme abortion restrictions on the books, with nearly all abortions banned in the state and an affirmative defense law that essentially asserts any doctor who provides an abortion is guilty until proven innocent. And now Idaho Republicans have set their sights on hindering certain residents from traveling out of state to get an abortion.
House Bill 242, which passed through the state House and is likely to move quickly through the Senate, seeks to limit minors’ ability to travel for abortion care without parental consent. The legislation would create a whole new crime — dubbed “abortion trafficking” — which is defined in the bill as an “adult who, with the intent to conceal an abortion from the parents or guardian of a pregnant, unemancipated minor, either procures an abortion … or obtains an abortion-inducing drug” for the minor. “Recruiting, harboring, or transporting the pregnant minor within this state commits the crime of abortion trafficking,” the legislation adds.
So a minor who has been raped and impregnated by a parent cannot get an abortion without parental consent and cannot be driven anywhere in the state to pick up an abortion drug, or the person driving them can be sentenced to two to five years in prison?
Read more on HuffPost.
Surveillance marketed as convenience? Is it really so difficult to use a credit card or (gasp) cash?
Biometric Technology: Jpmorgan Will Let Consumers Pay With Their Face Or Palm Instead Of A Card
JPMorgan Chase & Co. is planning to test new technology that would let consumers pay with their palms or faces at certain US merchants.
The bank, home to one of the world’s biggest payment-processing businesses, plans to roll out the service to its broader base of US merchant clients if the pilot program goes well, according to a statement Thursday. The pilot may include a Formula 1 race in Miami as well as some brick-and-mortar stores.
“The evolution of consumer technology has created new expectations for shoppers,” Jean-Marc Thienpont, head of omnichannel solutions for JPMorgan’s payments business, said in the statement. “Merchants need to be ready to adapt to these new expectations.
Drastic? Could we do this in the US? Or are we looking at too much money to ignore?
Meta bosses look at political ads ban in Europe
Meta executives are discussing a company-wide ban on political advertising in Europe, following concerns that its social networking platforms such as Facebook and Instagram will be unable to comply with forthcoming EU regulations that target online campaigning.
Brussels regulators are drawing up new laws to come into force next year designed to force large internet groups to reveal more about the political groups behind online campaigns and which users they are targeting.
Meta, led by chief executive Mark Zuckerberg, is concerned that the definition of political ads under the plan will be so broad that it will be easier to refuse all paid-for political campaigns on the company’s sites, according to two people briefed on internal discussions.
Perspective.
https://www.huntonprivacyblog.com/2023/03/29/uk-government-publishes-ai-white-paper/
UK Government Publishes AI White Paper
On March 29, 2023, the UK government published a white paper on artificial intelligence (“AI”) entitled “A pro-innovation approach to AI regulation.” The white paper sets out a new “flexible” approach to regulating artificial intelligence which is intended to build public trust in AI and make it easier for businesses to grow and create jobs.
No comments:
Post a Comment