Hamilton song written by AI features odd reference to Hillary Clinton

Hamilton song written by AI features odd reference to Hillary Clinton

Hard-core fans of the musical Hamilton can never get enough of the catchy soundtrack (I speak from experience). So they may be happy to know they now have a new song to earworm over. Lin-Manuel Miranda, the musical’s creator, had nothing to do with this one, though. The lyrics were written entirely by AI. 

To come up with the song’s lyrics, Eli Weiss, a film production student at California’s Chapman University, used Shortly Read, an AI application designed for writing that incorporates GPT-3, the powerful third-generation machine learning language model used by OpenAI,  a nonprofit artificial intelligence research group backed by Elon Musk. 

GPT-3 has been supplied with 45TB of text data, presumably including the full lyrics to Hamilton, and can generate a range of written content with simple inputs.  

Weiss and team entered this one sentence: “Here are the lyrics to a new song from the hit musical Hamilton: An American Musical.” The program then created lyrics to a tune with four verses, a chorus and a bridge that correctly identifies characters in the story and their relationships to each other. 

“It messes up a few times, like when Hillary Clinton makes a brief appearance,” says Weiss, a huge Hamilton fan, “but overall it’s incredibly convincing.”   

Indeed, most of the lyrics, both in words and cadence, feel like they’d fit right in to the musical, which tells the story of American Founding Father Alexander Hamilton, largely through hip-hop. “I wrote my way out of hell, I wrote 6 feet past the grave. I wrote a song about you, the only thing that kept me safe,” the AI lyrics go.   

GPT-3 produces word sequences that are often amazingly human-like, but can also contain some amusing surprises.    

In the case of the new Hamilton song, Hamilton’s devoted wife gives him a most unwelcome gift: “I met a certain young lady called Eliza and I’m 90% sure she gave me syphilis. But I hope I gave it back to you.” 

Then there’s the reference to Hillary Clinton as “my new Eliza.” The machine learning tool likely linked the former secretary of state to Hamilton lyrics referencing that government post. How Clinton becomes Hamilton’s new love is anyone’s AI guess. 

Weiss’ friend Michael Gribble, a film music student, put the AI-written song to music and performs it in the video above. This isn’t the first time AI has written a new Hamilton song, however. A few years ago, creative Max Deutsch trained a neural network on the musical’s lyrics and asked it to come up with a new tune. 

AI is becoming an increasingly visible player in the creative space, doing everything from generating Katy Perry and Elvis songs to painting nude portraits and crafting poetry in the style of the classics. Sometimes the results genuinely connect to the human experience. Other times, they’re downright creepy. 

Weiss and his creative partner Jacob Vaus are among those fascinated by AI’s creative potential and have tapped it to write scripts and compose other songs.

“Right now, most of that work still has this comedic charm to it, but I think somewhere in the middle of that spectrum is what we will start to see a lot of in the years to come,” Weiss says. “AI being used here and there in the creative process to fill in gaps and make adjustments.”

History has its eyes on you, AI. 

Published at Wed, 30 Dec 2020 01:59:34 +0000

Will use supervision tech to keep up with entities: RBI

TNN
Mumbai: The Reserve Bank of India (RBI) has said that it will start using SupTech (supervisory technology) and RegTech — a subset of fintech that uses technology to scale up supervision — to keep up with the entities it regulates. The central bank will soon conduct a survey on RegTech adoption and, based on the findings, will come out with broad principles to encourage adoption of these tools.
Currently, entities regulated by the RBI use technological tools like artificial intelligence, machine learning, big data analysis for know your customer & asset liability management (KYC/ALM) purposes, regulatory reporting & management information system, payments & account aggregation. Such tools are also used to judge the creditworthiness of borrowers.
“Notwithstanding its many advantages in terms of data and privacy protection, cyber risks are a major challenge in technology adoption,” the RBI said.
The use of technology for supervision has become crucial in light of the growing number of entities, including cooperative banks and finance companies, that are directly coming under the RBI’s lens. The central bank has indicated that on-site supervision will be largely for entities that are systemically important, while for most others it will depend on technology and analytics.
“Recognising that cutting-edge technology has enormous potential for preventive compliance, transaction monitoring and automated data flows, the RBI has accorded priority to adoption of RegTech,” the central bank said in a report. This will require regulated entities to submit information in machine-readable format.
Currently, the RBI’s offsite supervision depends heavily upon pre-defined templates to collect data. These are susceptible to inaccuracies and incomplete reporting. The RBI is trying to establish mechanisms to extract specific data directly from the source system to be more proactive in its risk-based supervision.
“The use of artificial intelligence and machine learning techniques are being explored to identify anomalies in the regulatory/supervisory reporting data, which can be used for predictive analysis. These techniques should pre-emptively help in micro-prudential supervision, identifying vulnerable branches, stressed exposures, unmitigated operational risks, suspicious transactions and misdemeanours,” the RBI said.
This kind of technology, where the regulator directly accesses data from the regulated entities’ system, is known as SupTech or supervisory technology.

Published at Tue, 29 Dec 2020 22:30:00 +0000