Thursday, January 9, 2025
Home Blog Page 1499

Stand-up comedians test ability of LLMs to write jokes

0
Stand-up comedians test ability of LLMs to write jokes


comedians use LLMs to write a stand-up routine
Credit: AI-generated image

A small team of AI researchers at Google’s DeepMind project has found that LLMs are not very good at writing jokes that are funny. They asked stand-up comedians to use LLMs to write a stand-up routine for them and posted their findings on the arXiv preprint server.

To be successful, most stand-up comedians have to write a stand-up routine and perform it on stage. Such routines, or monologs, typically involve both storytelling and jokes, or describe humorous situations. Many also employ surprising or incongruous remarks, giving the audience a sudden insight into something they may not have considered in a certain way before.

Most professional stand-up comedians spend a great deal of time polishing their routines and testing them on small audiences before performing in front of large crowds or on television specials.

In this new effort, the team at DeepMind wondered if LLMs might be capable of creating not just jokes, but entire stand-up routines. To find out, they recruited 20 professional stand-up comedians who had used LLMs in their work before. The performers used an LLM to help them write an entire routine and then rated the results.

The researchers found that LLMs were quite good at coming up with jokes; unfortunately, few if any were funny. Most, they suggested, were generic in nature and few offered anything in the way of a surprise.

Stand-up comedians test ability of LLMs to write jokes
Left: Evaluation of the instruction-tuned LLMs as creativity support tool for writing comedy using a Likert scale; each row corresponds to a question in a survey. Right box plots (box for quartiles and whiskers for min and max) show the break-down of the Creativity Support Index, respectively. Credit: arXiv (2024). DOI: 10.48550/arxiv.2405.20956

Overall, the comedians found the AI-generated jokes to lack the cutting edge that is typically needed for a joke to be funny. Many described the results as bland. But some of them did find the LLM to be useful in generating a routine that could be used for creating a basic structure around which they could build their own jokes.

The research team suggests the results were not surprising considering that makers of LLMs use filters to prevent them from generating output that could be offensive or edgy.

More information:
Piotr Wojciech Mirowski et al, A Robot Walks into a Bar: Can Language Models Serve as Creativity Support Tools for Comedy? An Evaluation of LLMs’ Humour Alignment with Comedians, arXiv (2024). DOI: 10.48550/arxiv.2405.20956

Journal information:
arXiv


© 2024 Science X Network

Citation:
Stand-up comedians test ability of LLMs to write jokes (2024, June 21)
retrieved 24 June 2024
from https://techxplore.com/news/2024-06-comedians-ability-llms.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Just believing that an AI is helping boosts your performance, study finds

0
Just believing that an AI is helping boosts your performance, study finds


Just believing that an AI is helping boosts your performance
The participants interacted with the system with their dominant hand using a keyboard and a mouse. Credit: Proceedings of the CHI Conference on Human Factors in Computing Systems (2024). DOI: 10.1145/3613904.3642633

Sometimes it seems like an AI is helping, but the benefit is actually a placebo effect—people performing better simply because they expect to be doing so—according to new research from Aalto University. The study also shows how difficult it is to shake people’s trust in the capabilities of AI systems.

In this study, participants were tasked with a simple letter recognition exercise. They performed the task once on their own and once supposedly aided by an AI system. Half of the participants were told the system was reliable and it would enhance their performance, and the other half was told that it was unreliable and would worsen their performance.

The findings are published in Proceedings of the CHI Conference on Human Factors in Computing Systems.

“In fact, neither AI system ever existed. Participants were led to believe an AI system was assisting them, when in reality, what the sham-AI was doing was completely random,” explains doctoral researcher Agnes Kloft.

The participants had to pair letters that popped up on screen at varying speeds. Surprisingly, both groups performed the exercise more efficiently—more quickly and attentively—when they believed an AI was involved.

“What we discovered is that people have extremely high expectations of these systems, and we can’t make them AI doomers simply by telling them a program doesn’t work,” says Assistant Professor Robin Welsch.

Following the initial experiments, the researchers conducted an online replication study that produced similar results. They also introduced a qualitative component, inviting participants to describe their expectations of performing with an AI. Most had a positive outlook toward AI and, surprisingly even skeptical people still had positive expectations about its performance.

The findings pose a problem for the methods generally used to evaluate emerging AI systems. “This is the big realization coming from our study—that it’s hard to evaluate programs that promise to help you because of this placebo effect,” Welsch says.

While powerful technologies like large language models undoubtedly streamline certain tasks, subtle differences between versions may be amplified or masked by the placebo effect—and this is effectively harnessed through marketing.

The results also pose a significant challenge for research on human-computer interaction, since expectations would influence the outcome unless placebo control studies were used.

“These results suggest that many studies in the field may have been skewed in favor of AI systems,” concludes Welsch.

More information:
Agnes Mercedes Kloft et al, “AI enhances our performance, I have no doubt this one will do the same”: The Placebo effect is robust to negative descriptions of AI, Proceedings of the CHI Conference on Human Factors in Computing Systems (2024). DOI: 10.1145/3613904.3642633

Provided by
Aalto University


Citation:
Just believing that an AI is helping boosts your performance, study finds (2024, May 13)
retrieved 24 June 2024
from https://techxplore.com/news/2024-05-believing-ai-boosts.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Apple holds talks with rival Meta over AI: Report

0
Apple holds talks with rival Meta over AI: Report


Apple
Credit: Armand Valendez from Pexels

Apple is talking to major rival Meta about integrating the Facebook parent company’s generative AI into its products, as it tries to catch up with rivals on artificial intelligence, the Wall Street Journal reported Sunday.

The report comes after Apple also struck a deal with OpenAI, the creator of ChatGPT, to help equip its Apple Intelligence suite of new AI features for its coveted products.

For months, pressure has been on Apple to persuade doubters on its AI strategy, after Microsoft and Google rolled out products in rapid-fire succession.

It has developed its own, smaller artificial intelligence but said that it will turn to others such as OpenAI to boost its in-house offering.

According to the Journal, which cited sources close to the matter, Meta has held discussions with Apple over integrating its own generative AI model into Apple Intelligence.

Apple senior vice president of software engineering Craig Federighi said in early June that Apple also wanted to integrate capabilities from Google’s generative AI system, Gemini, into its devices.

The big challenge for Apple has been how to infuse ChatGPT-style AI—which voraciously feeds off data—into its products without weakening its heavily promoted user privacy and security, according to analysts.

Apple Intelligence will enable users to create their own emojis based on a description in everyday language, or to generate brief summaries of e-mails in the mailbox.

Apple said Siri, its voice assistant, will also get an AI-infused upgrade and now will appear as a pulsating light on the edge of your home screen.

Launched over 12 years ago, Siri has long since been seen as a dated feature, overtaken by the new generation of assistants, such as GPT-4o, OpenAI’s latest offering.

According to Canalys, 16 percent of smartphones shipped this year will be equipped with generative AI features, a proportion it expects to rise to 54 percent by 2028.

© 2024 AFP

Citation:
Apple holds talks with rival Meta over AI: Report (2024, June 24)
retrieved 24 June 2024
from https://techxplore.com/news/2024-06-apple-rival-meta-ai.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Wirelessly-powered relay will help bring 5G technology to smart factories, say researchers

0
Wirelessly-powered relay will help bring 5G technology to smart factories, say researchers


Wirelessly powered relay will help bring 5G technology to smart factories
A key innovation in this design is the use of the 5.7 GHz wireless power transfer (WPT) signal as both a means of generating DC power using a rectifier and as an oscillator for the mixing and unmixing circuits. By amplifying the input signal after down conversion to a lower frequency via mixing, this circuit achieves higher efficiency and gain. Credit: 2024 IEEE Symposium on VLSI Technology & Circuits

A recently-developed wirelessly-powered 5G relay could accelerate the development of smart factories, report scientists from Tokyo Tech. By adopting a lower operating frequency for wireless power transfer, the proposed relay design solves many of the current limitations, including range and efficiency. In turn, this allows for a more versatile and widespread arrangement of sensors and transceivers in industrial settings.

One of the hallmarks of the Information Age is the transformation of industries towards a greater flow of information. This can be readily seen in high-tech factories and warehouses, where wireless sensors and transceivers are installed in robots, production machinery, and automatic vehicles. In many cases, 5G networks are used to orchestrate operations and communications between these devices.

To avoid relying on cumbersome wired power sources, sensors and transceivers can be energized remotely via wireless power transfer (WPT). However, one problem with conventional WPT designs is that they operate at 24 GHz.

At such high frequencies, transmission beams must be extremely narrow to avoid energy losses. Moreover, power can only be transmitted if there is a clear line of sight between the WPT system and the target device. Since 5G relays are often used to extend the range of 5G base stations, WPT needs to reach even further, which is yet another challenge for 24 GHz systems.

To address the limitations of WPT, a research team from Tokyo Institute of Technology has come up with a clever solution. In a recent study, presented at the 2024 IEEE Symposium on VLSI Technology & Circuits, they developed a novel 5G relay that can be powered wirelessly at a lower frequency of 5.7 GHz.

“By using 5.7 GHz as the WPT frequency, we can get wider coverage than conventional 24 GHz WPT systems, enabling a wider range of devices to operate simultaneously,” explains senior author and Associate Professor Atsushi Shirane.

Wirelessly powered relay will help bring 5G technology to smart factories
The prototype of the proposed relay transceiver was fabricated with Si CMOS 65nm chips and 4×2 patch phased-array antenna board. Credit: 2024 IEEE Symposium on VLSI Technology & Circuits

The proposed wirelessly-powered relay is meant to act as an intermediary receiver and transmitter of 5G signals, which can originate from a 5G base station or wireless devices. The key innovation of this system is the use of a rectifier-type mixer, which performs 4th-order subharmonic mixing while also generating DC power.

Notably, the mixer uses the received 5.7 GHz WPT signal as a local signal. With this local signal, together with multiplying circuits, phase shifters, and a power combiner, the mixer ‘down-converts’ a received 28 GHz signal into a 5.2 GHz signal. Then, this 5.2 GHz signal is internally amplified, up-converted to 28 GHz through the inverse process, and retransmitted to its intended destination.

To drive these internal amplifiers, the proposed system first rectifies the 5.7 GHz WPT signal to produce DC power, which is managed by a dedicated power management unit. This ingenious approach offers several advantages, as Shirane highlights, “Since the 5.7 GHz WPT signal has less path loss than the 24 GHz signal, more power can be obtained from a rectifier. In addition, the 5.7 GHz rectifier has a lower loss than 24 GHz rectifiers and can operate at a higher power conversion efficiency.”

Finally, this proposed circuit design allows for selecting the transistor size, bias voltage, matching, cutoff frequency of the filter, and load to maximize conversion efficiency and conversion gain simultaneously.

Through several experiments, the research team showcased the capabilities of their proposed relay. Occupying only a 1.5 mm by 0.77 mm chip using standard CMOS technology, a single chip can output a high power of 6.45 mW at an input power of 10.7 dBm. Notably, multiple chips could be combined to achieve a higher power output. Considering its many advantages, the proposed 5.7 GHz WPT system could thus greatly contribute to the development of smart factories.

More information:
Presentation: A 28GHz 5GNR Wirelessly Powered Relay Transceiver Using Rectifier-Type 4th-Order Sub-Harmonic Mixer

Citation:
Wirelessly-powered relay will help bring 5G technology to smart factories, say researchers (2024, June 17)
retrieved 24 June 2024
from https://techxplore.com/news/2024-06-wirelessly-powered-relay-5g-technology.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link

Hidden humor, the software developer’s secret weapon

0
Hidden humor, the software developer's secret weapon


coder
Credit: Pixabay/CC0 Public Domain

Writing software code can be a painstaking and stressful process—and downright boring when the job is repetitive and you’re doing it remotely, alone in front of your screen.

To liven things up, many developers and testers use humor to relieve the monotony and connect with their virtual colleagues by sharing a joke. Over time, it creates a bond with fellow developers, though the humor and creativity slipped in between the lines of code are invisible to the rest of us.

“Humor creates relationships between people who are physically distant and is a good way to stave off boredom,” said Benoit Baudry, a professor in the Department of Computer Science and Operations Research at Université de Montréal. “It’s a way to build engagement.” Until recently, Baudry was at the Royal Institute of Technology in Stockholm, where he and his colleagues studied the special humor of developers.

“Developers are people who love software,” said Baudry. “So they try to create emotional bonds using the digital technology that is their work tool.”

But they have to exercise some caution about when and where they insert jokes and comments. They don’t want any of their jests to end up on Instagram.

To find out more about how they do it, Baudry and his fellow researchers circulated an online questionnaire that was posted on developer sites. More than 125 developers from around the world responded. They reported using humor most frequently in test inputs and “commits,” or changes to the code. A sly dialogue unfolds between the test lines.

The research was published in the Proceedings of the 46th International Conference on Software Engineering: Software Engineering in Society on June 6 and is also available on the arXiv preprint server.

Darth Vader, Luke et al

Baudry and his co-authors looked at Faker, a library that generates random data for use in testing code. Instead of lorem ipsum—a sequence of meaningless words commonly used as a placeholder for text on a page until it can be replaced by the real thing—developers will sprinkle their lines with cultural references such as allusions to Seinfeld or quotes from poets.

“Some references are fairly specific, others are universal: who doesn’t know the characters from Star Wars or The Matrix?” said Baudry. Characters from cult films are frequently used in titles, as are quotations. An example from Faker: “The wise animal blends into its surroundings” (a quote from the movies Dune). Nothing edgy or inappropriate, just light-hearted asides through which developers signal their interests and elicit a smile from their colleagues.

“Personally, I like to use characters from the 1998 film The Big Lebowski in error messages,” said Baudry. He is also enthusiastic about lolcommits, a utility that lets developers send a selfie when they make changes to code. “These pics foster bonds with colleagues and are a way to celebrate when the job is done,” he said.

The pioneer who paved the way

The trailblazer for quips in code was the brilliant NASA engineer and computer scientist Margaret Hamilton, who led the team that designed the system for the Apollo 11 lunar landing program in 1969. When the code was made public, people could see that it was peppered with jokes, Shakespeare quotes and references to The Wizard of Oz.

Humor in code “helps keep it fun,” one of the respondents to the survey commented. “I love it and think fondly of people writing that part of the code or comment.”

Humor “makes a codebase feel more humanized, like it was created by a real person,” another respondent said.

Naturally, there are limits to the kind of humor that can be injected into code. “It should not create a toxic or unwelcoming culture,” cautioned one respondent.

Baudry’s interest in tech humor is not new. Last year, he published a fascinating article on “Easter eggs,” features hidden in software which can be unlocked by pressing a combination of keys or correctly positioning the pointer. But unlike code humor, Easter eggs can be discovered by the public, especially in video games.

Baudry also wants the users of technology to be more aware of the behind-the-scenes human activity that produces the thousands of connections and apps that are woven into our lives. In the past, he has given talks on art and technology while projecting code onto giant screens in public places. For the love of code!

More information:
Deepika Tiwari et al, With Great Humor Comes Great Developer Engagement, Proceedings of the 46th International Conference on Software Engineering: Software Engineering in Society (2024). DOI: 10.1145/3639475.3640099. On arXiv: DOI: 10.48550/arxiv.2312.01680

Journal information:
arXiv


Citation:
Hidden humor, the software developer’s secret weapon (2024, June 21)
retrieved 24 June 2024
from https://techxplore.com/news/2024-06-hidden-humor-software-secret-weapon.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no
part may be reproduced without the written permission. The content is provided for information purposes only.





Source link