Tag: Artificial Intelligence

AI models show promise in predicting heart disease risks, but lack validation
[ad_1]
In a recent review published in BMC Medicine, scientists evaluate artificial intelligence models (AI-Ms) that predict cardiovascular disease (CVD) risks in general and specific populations while also developing an independent validation score (IVS) for AI-Ms.

Study: Artificial intelligence in the risk prediction models of cardiovascular disease and development of an independent validation screening tool: a systematic review. Image Credit: Summit Art Creations / Shutterstock.com

Background

The global prevalence of cardiovascular diseases (CVDs) is increasing rapidly, which has led to the development of several CVD prediction models. CVD prediction models like Framingham and SCORE identify individuals at a greater risk of developing CVDs to ultimately implement preventive measures across the vulnerable population.

Within computer science, the application of AI, machine learning (ML), and deep learning (DL) can be used to develop computational systems with a similar functioning capacity analogous to human intelligence while performing a complex task. This functioning capacity is associated with humans’ reasoning, learning, perception, problem-solving, decision-making, and language comprehension skills.

AI-Ms have been increasingly applied in the healthcare sector for disease risk prediction. However, this application has been subjected to multiple challenges linked to data privacy, security, transparency, legality, and concerns related to ethics. Nevertheless, as compared to traditional risk prediction models, AI-Ms are associated with greater accuracy, data-processing capability, and fewer processing restrictions.

About the study

Extensive data extraction was performed based on predictors, algorithms, bias, and population. A tool to assess the replicability and applicability of AI-Ms, as well as ensure the external validation of AI-Ms, was developed to screen AI-Ms.

For the current review, all relevant articles were obtained from Embase, Web of Science, PubMed, and IEEE Library. The prediction risk of bias assessment tool (PROBAST) was also used.

Key findings

A total of 79 relevant articles published between 2017 and 2021 were obtained, of which 486 AI-Ms were identified. Most of these studies were related to the development of new AI-Ms; however, none of the models underwent independent external validation.

Thus, AI risk prediction researchers appear to be more focused on developing new models than validating existing ones, which is crucial for clinical applications. Since unvalidated AI-Ms would result in the generation of many useless prediction models, researchers must focus on validating AI-Ms to avoid wasting research time.

A key factor that restricts the implementation of external validation is the use of limited data sources for model development. However, this could be addressed by using data from multi-source databases.

Most AI-based models as CVD risk predictors were developed in North America and Europe, very few of which were developed in Asian and South American countries, whereas none were developed in Africa. Since the extent of CVD risks varies among ethnicities, it is important to develop AI-Ms that focus on specific ethnic groups.

The four most common variables used in AI-Ms for CVD risk predictions include total cholesterol, age, sex, and smoking status. Compared to traditional models, AI-Ms evaluate multimodal data, including additional gene- or protein-related information and image data. Other advantages of AI models include data re-input and utility.

Many studies did not provide important research information, which compromised model validation. In the future, studies must provide a Transparent Reporting of a multivariable prediction model for the Individual Prognosis Or Diagnosis (TRIPOD) statement when the manuscript is submitted.

According to PROBAST, all models were at a high risk of bias, primarily because of the inappropriate use of statistical tools. IVS analysis revealed that only 10 models were “recommended” for use, whereas the remaining models were categorized under “not recommended” or “warning.”

The IVS tool has been developed for screening independent external validation models. This scoring system evaluates the suitability for independent external validation based on transparency, risk assessment, performance, and clinical implication.

The newly developed IVS indicated that independent external validation may not be suitable for over 95% of the models, thus implying that these models cannot be used in clinical settings.

Conclusions

Although several AI-Ms for CVD predictions are available, few studies have systematically analyzed the models for their effectiveness. The current review summarized AI-Ms for CVD and discussed current challenges associated with their use.

The current study provided important insights into AI models used for CVD risk predictions, including the geographical imbalance, a high risk of bias, a low standard-reaching rate of report quality, a lack of independent external validation, and an imperfect evaluation system. In this context, the use of a newly developed IVS tool could help assess the replicability of the models.
Journal reference:

Cai, Y., Ca, Y., Tang, L., et al. (2024) Artificial intelligence in the risk prediction models of cardiovascular disease and development of an independent validation screening tool: a systematic review. BMC Medicine 56. doi:10.1186/s12916-024-03273-7
[ad_2]

Source link
February 9, 2024
AI’s ability to detect COVID-19 from coughs faces real-world challenges
[ad_1]
A recent Nature Machine Intelligence study investigated the efficacy of audio-based artificial intelligence (AI) classifiers in predicting severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection status. SARS-CoV-2 is the causal organism of the coronavirus disease 2019 (COVID-19) pandemic.

Study: Audio-based AI classifiers show no evidence of improved COVID-19 screening over simple symptoms checkers. Image Credit: Aliaksandra Post / Shutterstock

Background

Since SARS-CoV-2 infection could cause both symptomatic and asymptomatic manifestations, it is important to develop accurate tests to avoid general population quarantine. Previous studies have revealed that AI-based classifiers trained with respiratory audio data could identify SARS-CoV-2 status.

Although these studies indicated the effectiveness of AI-based classifiers, many challenges surfaced while applying them in real-world settings. Some factors that withheld AI-based classifier applications were sampling biases, unvalidated data on participants’ COVID-19 status, and delay between infection and audio recording. It is imperative to determine whether the audio biomarkers of COVID-19 are unique to SARS-CoV-2 infection or are inappropriate confounding signals.

About the Study

The current study focussed on determining whether audio-based classifiers can be accurately used for COVID-19 screening. A large-scale polymerase chain reaction (PCR) dataset linked to audio-based COVID-19 screening (ABCS) was used. For this study, participants of the Real-time Assessment of Community Transmission (REACT) program and the National Health Service (NHS) Test-and-Trace (T+T) service were invited. All relevant demographic data was extracted from T+T/REACT records.

Participants were asked to complete survey questions and record four audio clips. For audio recordings, they were asked to read a specific sentence, followed by three successive exhalations, making a “ha” sound. Furthermore, the participants were asked to record forced coughs once and three times in succession. All recordings were documented in .wav format. The quality of the audio recordings was assessed, and 5,157 records were removed for quality-related issues.

Human figures represent study participants and their corresponding COVID-19 infection status, with the different colours portraying different demographic or symptomatic features. When participants are randomly split into training and test sets, the randomized split models perform well at COVID-19 detection, achieving AUCs in excess of 0.8; however, matched test set performance is seen to drop to estimated AUC between 0.60 and 0.65, with an AUC of 0.5 representing random classification. Inflated classification performance is also seen in engineered out of distribution test sets such as: the designed test set, in which a select set of demographic groups appear solely in the testing set, and the longitudinal test set, in which there is no overlap in the time of submission between train and test instances. The 95% confidence intervals calculated via the normal approximation method are shown, along with the corresponding n numbers of the train and test sets.

Study Findings

In this study, a respiratory acoustic dataset of 67,842 individuals was collected. Among them, 23,514 individuals tested positive for COVID-19. All data were linked with PCR test results. It must be noted that the most significant number of COVID-19-negative participants were recruited from six REACT rounds compared to the T+T channel.

The dataset considered in this study exhibited promising coverage across England. No significant association between geographical location and COVID-19 status was noted. The highest level of COVID-19 imbalance was found in Cornwall. A previous study indicated recruitment bias in ABCS, particularly linked with age, language, and gender, in both training data and test sets. Despite this bias, the training dataset was balanced in accordance with age and gender across COVID-positive and COVID-negative subgroups.

Consistent with previous studies, the unadjusted analysis conducted in this study exhibited that AI classifiers can predict COVID-19 status with high accuracy. However, when measured confounders were matched, a weak performance of AI classifiers in detecting SARS-CoV-2 status was observed.

Based on the findings, the current study proposed some guidelines to rectify recruitment bias’s effect for future studies. Some of the recommendations are listed below:
1. Audio samples stored in repositories must include details of the study recruitment criteria. In addition, relevant information about the individuals, including their gender, age, time of COVID-19 test, SARS-CoV-2 symptoms, and locations, must be documented along with the audio recording.
2. All confounding factors must be identified and matched to help control recruitment bias.
3. Experimental design must be developed, keeping the possible bias in mind. In most cases, data matching leads to a reduction in sample size. Observational studies recruit participants focusing on the maximized possibility of matching measured confounders.
4. The predictive values of the classifiers must be compared with standard protocol findings.
5. AI classifiers’ predictive accuracy must be assessed. However, the predictive accuracy, sensitivity, and specificity vary depending on the targeted population.
6. The classifiers’ utility must be assessed for each testing outcome.
7. The replication study must be conducted in randomized cohorts. Furthermore, pilot studies must be conducted in real-world settings based on domain-specific utility.
Conclusions

The current study has come with limitations that include the possibility of potential unmeasured confounders across REACT and T+T recruitment channels. For instance, PCR testing for COVID-19 was performed several days after self-screening of symptoms. In contrast, PCR tests in REACT were conducted on a pre-determined date, irrespective of the onset of symptoms. Although the majority of confounders were matched, there is a possibility of the presence of residual predictive variation.

Despite the limitations, this study highlighted the need to develop accurate machine-learning evaluation procedures to obtain unbiased outputs. Furthermore, it revealed that confounding factors are hard to detect and control across many AI applications.
Journal reference:

Coppock, H. et al. (2024) Audio-based AI classifiers show no evidence of improved COVID-19 screening over simple symptoms checkers. Nature Machine Intelligence. 1-14. DOI: 10.1038/s42256-023-00773-8, https://www.nature.com/articles/s42256-023-00773-8
[ad_2]

Source link
February 9, 2024
AI Tools Like GitHub Copilot Are Rewiring Coders’ Brains. Yours May Be Next

[ad_1]

Many people—like, say, journalists—are understandably antsy about what generative artificial intelligence might mean for the future of their profession. It doesn’t help that expert prognostications on the matter offer a confusing cocktail of wide-eyed excitement, trenchant skepticism, and dystopian despair.

Some workers are already living in one potential version of the generative AI future, though: computer programmers.

“Developers have arrived in the age of AI,” says Thomas Dohmke, CEO of GitHub. “The only question is, how fast do you get on board? Or are you going to be stuck in the past, on the wrong side of the ‘productivity polarity’?”

In June 2021, GitHub launched a preview version of a programming aid called Copilot, which uses generative AI to suggest how to complete large chunks of code as soon as a person starts typing. Copilot is now a paid tool and a smash hit. GitHub’s owner, Microsoft, said in its latest quarterly earnings that there are now 1.3 million paid Copilot accounts—a 30 percent increase over the previous quarter—and noted that 50,000 different companies use the software.

Dohmke says the latest usage data from Copilot shows that almost half of all the code produced by users is AI-generated. At the same time, he claims there is little sign that these AI programs can operate without human oversight. “There’s clear consensus from the developer community after using these tools that it needs to be a pair-programmer copilot,” Dohmke says.

Copilot’s power is in how it abstracts away complexity for a programmer trying to work through a problem, Dohmke says. He likens that to the way modern programming languages hide fiddly details that earlier, lower-level languages required coders to wrangle. Dohmke adds that younger programmers are particularly accepting of Copilot, and that it seems especially helpful in solving novice coding problems. (This makes sense if you consider that Copilot learned from reams of code posted online, where solutions to beginner problems outnumber examples of abstruse and rarified coding craft.)

“We’re seeing the evolution of software development,” Dohmke says.

None of that means demand for developers’ labor won’t be altered by AI. GitHub research in collaboration with MIT shows that Copilot allowed coders faced with relatively simple tasks to complete their work, on average, 55 percent more quickly. This increase in productivity suggests that companies could get the same work done with fewer programmers, but companies could use those savings to spend more on labor in other projects.

Even for non-coders, these findings—and the rapid uptake of Copilot—are potentially instructive. Microsoft is developing AI Copilots, as it calls them, designed to help write emails, craft spreadsheets, or analyze documents for its Office software. It even introduced a Copilot key to the latest Windows PCs, its first major keyboard button change in decades. Competitors like Google are building similar tools. GitHub’s success might be helping to drive this push to give everyone an AI workplace assistant.

“There’s good empirical evidence and data around the GitHub Copilot and the productivity stats around it,” Microsoft’s CEO, Satya Nadella, said on the company’s most recent earnings call. He added that he expects similar gains to be felt among users of Microsoft’s other Copilots. Microsoft has created a site where you can try its Copilot for Windows. I confess it isn’t clear to me how similar the tasks you might want to do on Windows are to the ones you do in GitHub Copilot, where you use code to achieve clear objectives.

There are other potential side effects of tools like GitHub Copilot besides job displacement. For example, increased reliance on automation might lead to more errors creeping into code. One recent study claimed to find evidence of such a trend—although Dohmke says that it reported only a general increase in mistakes since Copilot was introduced, not direct evidence that the AI helper was causing an increase in errors. While this is true, it seems fair to worry that less experienced coders might miss errors when relying on AI help, or that the overall quality of code might decrease thanks to autocomplete.

Given Copilot’s popularity, it won’t be long before we have more data on that question. Those of us who work in other jobs may soon find out whether we’re in for the same productivity gains as coders—and the corporate upheavals that come with them.

[ad_2]

Source link

February 8, 2024
London Underground Is Testing Real-Time AI Surveillance Tools to Spot Crime

[ad_1]

Staff at the transportation body ran “extensive simulations” at Willesden Green station during the trial to gather more training data, the documents say. These included members of staff falling on the floor, and some of these tests happened when the station was closed. “You will see the BTP [British Transport Police] officer holding a machete and handgun in different locations within the station,” one caption in the documents state, although the images are redacted. During the trial, the files say, there were no alerts for weapons incidents at the station.

The most alerts were issued for people potentially avoiding paying for their journeys by jumping over or crawling under closed fare gates, pushing gates open, walking through open gates, or tailgating someone who paid. Fare dodging costs up to £130 million per year, TfL says, and there were 26,000 fare evasion alerts during the trial.

During all of the tests, images of people’s faces were blurred and data was kept for a maximum of 14 days. However, six months into the trial, the TfL decided to unblur the images of faces when people were suspected of not paying, and it kept that data for longer. It was originally planned, the documents say, for staff to respond to the fare dodging alerts. “However, due to the large number of daily alerts (in some days over 300) and the high accuracy in detections, we configured the system to auto-acknowledge the alerts,” the documents say.

Birtwistle, from the Ada Lovelace Institute, says that people expect “robust oversight and governance” when technologies like these are put in place. “If these technologies are going to be used, they should only be used with public trust, consent and support,” Birtwistle says.

A large part of the trial was aimed at helping staff understand what was happening at the station and respond to incidents. The 59 wheelchair alerts allowed staff at Willesden Green station, which does not have access facilities for wheelchairs, to “provide the necessary care and assistance,” the files say. Meanwhile, there were almost 2,200 alerts for people going beyond yellow safety lines, 39 for people leaning over the edge of the track, and almost 2,000 alerts for people sitting on a bench for extended periods.

“Throughout the PoC we have seen a huge increase in the number of public announcements made by staff, reminding customers to step away from the yellow line,” the documents say. They also say the system generated alerts for “rough sleepers and beggars” at the station’s entrances and claim this allowed staff to “remotely monitor the situation and provide the necessary care and assistance.” TfL states that the system was trialed to try to help it improve the quality of staffing at its stations and make it safer for passengers.

The files do not contain any analysis of how accurate the AI detection system is; however, at various points, the detection had to be adjusted. “Object detection and behavior detection are generally quite fragile and are not foolproof,” Leufer, of Access Nows, says. In one instance, the system created alerts saying people were in an unauthorized area when in reality train drivers were leaving the train. Sunlight shining onto the camera also made them less effective, the documents say.

[ad_2]

Source link

February 8, 2024
AI-Generated Voices in Robocalls Are Now Illegal

[ad_1]

It’s now illegal for robocallers to use AI-generated voices thanks to a new ruling by the Federal Communications Commission on Thursday.

In a unanimous decision, the FCC expands the Telephone Consumer Protection Act, or TCPA, to cover robocall scams that contain AI voice clones. The new rule goes into effect immediately, allowing for the commission to fine companies and block providers for making these types of calls.

“Bad actors are using AI-generated voices in unsolicited robocalls to extort vulnerable family members, imitate celebrities, and misinform voters,” FCC Chairwoman Jessica Rosenworcel said in a statement on Thursday. “We’re putting the fraudsters behind these robocalls on notice.”

The move comes a few days after the FCC and New Hampshire Attorney General John Formella identified Life Corporation as the company behind the mysterious robocalls imitating President Joe Biden last month before the state’s primary election. At a Tuesday press conference, Formella said that his office had opened a criminal investigation into the company and its owner, Walter Monk.

The FCC first announced its plan to outlaw AI-generated robocall scams by updating the TCPA last week. The agency has used the law in the past to go after junk callers, including the conservative activists and pranksters Jacob Wohl and Jack Burkman. In 2021, the FCC fined them more than $5 million for conducting a massive robocalling scheme to discourage voters from voting by mail in the 2020 election.

“While this generative AI technology is new, and it poses a lot of challenges, we already have a lot of the tools that we need to grapple with that challenge,” Nicholas Garcia, policy counsel at Public Knowledge, tells WIRED. “We can apply existing laws like the TCPA and a regulatory agency like the FCC has the flexibility and the expertise to go in and respond to these threats in real time.”

[ad_2]

Source link

February 8, 2024
How to Get Gemini Advanced, Google’s Subscription-Only AI Chatbot

[ad_1]

Google just upgraded its AI chatbot. Access to the new Gemini Advanced chatbot is available right now through a monthly subscription to Google One, the company’s cloud backup service.

With today’s release, there’s a new tier of Google One service called the AI Premium plan, which costs $20 per month; subscribe and you’ll get the standard Google One benefits, plus access to the new chatbot.

That $20 monthly fee is the same price OpenAI charges for access to its GPT-4 model via ChatGPT Plus. If you don’t want to pay for your AI tools, Google also offers a more basic option called Gemini Pro that remains free to use. Google previously offered a chatbot known as “Bard,” but the company has renamed its AI offerings as “Gemini.”

While aspects of Google’s Gemini Ultra model were shown off months ago in sometimes questionable demos, the juiced up version was not made available to the public during Gemini’s initial rollout. Now the more capable version has arrived.

Gemini Advanced, similar to OpenAI’s GPT-4, may be better at understanding the nuance and context of user prompts. It should also perform better at jobs like writing code when compared to previous large language models. Gemini Advanced is currently just designed for answering prompts in English, but additional languages are expected to receive support with future iterations.

Curious about trying out Google’s latest chatbot and wondering how the subscription plan compares to ChatGPT Plus? Here’s everything you need to know about getting started with Gemini Advanced.

How to Access Google’s Gemini Advanced

Instead of offering access through a stand-alone subscription, Google is making Gemini Advanced available as part of a new tier for Google One called “AI Premium.” If you don’t already have a Google account, you’ll need to create one before signing up to use the chatbot. If you do already subscribe to a Google One plan, you’ll be able to upgrade to the new plan that includes chatbot access. Check out WIRED’s guide to Google One for a deeper dive into what’s included with the different tiers.

After you sign up for the $20 monthly AI Premium plan—Google offers a two-month free trial if you’re unsure—log in to the Gemini chatbot in your web browser. The option to use Gemini Advanced should appear. (With the rebranding of Bard comes a different website URL for Gemini, so make sure to update any old browser bookmarks.) The new AI model is also available on your smartphone. Download the fresh Gemini app for Android, or find the new chatbot inside the standard Google app for iOS devices.

Courtesy of Google

How does the subscription compare to what’s offered by OpenAI? With ChatGPT Plus, you get to use the impressive GPT-4 model and try out ChatGPT’s beta features before they are released to a wider audience. With Gemini Advanced, you get access to the Google’s latest AI model and new features as well as everything included with the Google One subscription. For $20 a month, you receive 2 terabytes of cloud storage, access to advanced photo editing features like Magic Eraser, and 10 percent back in credit for hardware purchases from the Google Store.

The company plans to roll out a Gemini integration with Gmail and Google Docs for AI Premium subscribers, but the exact release date for this feature is unknown. When WIRED previously tested Bard’s tricks in Gmail, the initial results were messy but showed promise. It will be interesting to see if a more powerful large language model improves the user experience of trying to find old emails in an overstuffed inbox.

[ad_2]

Source link

February 8, 2024
Gemini Advanced Is a Central Part of Google’s Subscription Future

[ad_1]

Google got to where it is mostly by offering free services stuffed with ads, but it has increasingly experimented with a different business model: selling subscriptions for extra perks. Its first subscription offering, debuted in 2006, provided additional photo storage for users who didn’t want to have to hit the delete button. You can now pay Google for extra space for emails and documents, too, or to keep recordings from Nest security cameras and remove ads from YouTube. Today the company added a major new pitch to its subscription slate—it’s asking people to pay extra to access a smarter AI chatbot and more capable productivity helpers.

Gemini Advanced, Google’s most powerful chatbot yet, launched today behind a paywall. It costs $19.99 a month in a new tier of the Google One subscription plan known as AI Premium. It combines access to the new chatbot with existing Google One offerings like 2 terabytes of extra storage, a VPN, and other perks.

AI Premium is similar in price to OpenAI’s $20 a month ChatGPT Plus, but includes Google One benefits that otherwise cost $9.99 a month. Subscribers already on pricier Google One tiers will get the new Gemini Advanced features through July 31 at no extra cost; it’s unclear what happens after then.

Google has said Gemini is at the heart of its plans for an AI-enhanced future. If AI Premium finds an audience, that future could also include Google drawing a significant new revenue stream from subscriptions, as people pay to access more powerful AI tools much as gamers shell out for more powerful hardware.

Convincing consumers to cough up for AI could also be imperative for Google. Though hard drive costs generally keep falling, prices for powerful chips such as the Nvidia GPUs and Google TPUs needed for cutting-edge generative AI projects are shockingly high as demand outpaces supply.

Shimrit Ben-Yair, vice president and general manager for the Google One business, tells WIRED that defraying the costs of the computing power behind Gemini is “definitely part of the thinking” in requiring a subscription to access the most advanced version. And it won’t be the last time Google launches an AI feature behind a paywall. “It’s just the first step in many more of these generative AI features coming to the market through Google One,” she says.

Google announced last week that Google One was about to cross 100 million subscribers. Ben-Yair says that AI Premium is central to how Google expects to land its next 100 million.

New Model

The generative AI chip crunch also explains why AI Premium comes with a significant restriction despite being an expansion of Google One: While established Google One benefits such as storage can be split among six Google accounts without additional fees, only the plan manager will have access to Gemini Advanced. “We want to build a really sustainable long-term business here,” Shimrit Ben-Yair says.

In an interview with WIRED about Google’s AI strategy in the Gemini era, CEO Sundar Pichai says the company projected costs and potential efficiencies 25 years out when determining pricing for the AI Premium plan. The aim was to keep fees, in the company’s view, compelling, while also providing cash to support new development. “We’re building it in a way so that over time it’s what will allow us to invest more in the models and create that virtuous cycle,” Pichai says.

[ad_2]

Source link

February 8, 2024
Google Rebrands Its AI Chatbot as Gemini to Take On ChatGPT

[ad_1]

When OpenAI’s ChatGPT opened a new era in tech, the industry’s former AI champ, Google, responded by reorganizing its labs and launching a profusion of sometimes overlapping AI services. This included the Bard chatbot, workplace helper Duet AI, and a chatbot-style version of search.

Now Google is consolidating many of its generative AI products under the banner of its latest AI model Gemini—and taking direct aim at OpenAI’s subscription service ChatGPT Plus.

Google announced today that Bard, its experimental chatbot hurriedly launched last March, is now called Gemini—taking the same name of the text, voice, and image capable AI model that started powering the Bard chatbot back in December. Gemini is also getting more prominent positioning among Google’s services. It will have its own app on Android phones, and on Apple mobile devices Gemini will be baked into the primary Google app.

When Google first unveiled the Gemini AI model it was portrayed as a new foundation for its AI offerings, but the company had held back the most powerful version, saying it needed more testing for safety. That version, Gemini Ultra, is now being made available inside a premium version of Google’s chatbot, called Gemini Advanced. Accessing it requires a subscription to a new tier of the Google One cloud backup service called AI Premium. Typically, a $10 subscription to Google One comes with 2 terabytes of extra storage and other benefits; now that same package is available with Gemini Advanced thrown in for $20 per month.

That new bundle from Google offers significantly more than a subscription to OpenAI’s ChatGPT Plus, which costs $20 a month. The service includes access to the company’s most powerful version of its chatbot and also OpenAI’s new “GPT store”, which offers custom chatbot functions crafted by developers. For the same monthly cost, Google One customers can now get extra Gmail, Drive and Photo storage in addition to a more powerful chat-ified search experience.

Personality Upgrade

Sissie Hsiao, a VP at Google and general manager for Google Assistant and Bard, said in a media briefing ahead of today’s launch that Google conducted blind tests with users of Gemini and other leading chatbots and found the Google offering to be “the most preferred chatbot.”

Hsiao said Google also gave a hundred leading AI experts access to the advanced version of Gemini and asked them to challenge the bot with complex requests. “They’ve been really excited and giving us really positive feedback.”

Google says the new Gemini will now have more attitude—a departure from the more neutral tone that it previously adopted—and will “understand intent and react with personality,” according to Jack Krawczyk, a Google director of product management. That may be inspired by the downright ebullient chatbots launched by some smaller AI upstarts, such as Pi from startup Inflection AI, and the various app-specific personas that ChatGPT’s custom GPTs now have.

[ad_2]

Source link

February 8, 2024
Google Prepares for a Future Where Search Isn’t King

[ad_1]

Google’s CEO Sundar Pichai still loves the web. He wakes up every morning and reads Techmeme, a news aggregator resplendent with links, accessible only via the web. The web is dynamic and resilient, he says, and can still—with help from a search engine—provide whatever information a person is looking for.

Yet the web and its critical search layer are changing. We can all see it happening: Social media apps, short-form video, and generative AI are challenging our outdated ideals of what it means to find information online. Quality information online. Pichai sees it, too. But he has more power than most to steer it.

The way Pichai is rolling out Gemini, Google’s most powerful AI model yet, suggests that much as he likes the good ol’ web, he’s much more interested in a futuristic version of it. He has to be: The chatbots are coming for him.

Today Google announced that the chatbot it launched to counter OpenAI’s ChatGPT, Bard, is getting a new name: Gemini, like the AI model it’s based on that was first unveiled in December. The Gemini chatbot is also going mobile, and inching away from its “experimental” phase and closer to general availability. It will have its own app on Android and prime placement in the Google search app on iOS. And the most advanced version of Gemini will also be offered as part of a $20 per month Google One subscription package.

In releasing the most powerful version of Gemini with a paywall, Google is taking direct aim at the fast-ascendant ChatGPT and the subscription service ChatGPT Plus. Pichai is also experimenting with a new vision for what Google offers—not replacing search, not yet, but building an alternative to see what sticks.

“This is how we’ve always approached search, in the sense that as search evolved, as mobile came in and user interactions changed, we adapted to it,” Pichai says, speaking with WIRED ahead of the Gemini launch. “In some cases we’re leading users, as we are with multimodal AI. But I want to be flexible about the future, because otherwise we’ll get it wrong.”

Sensory Overload

“Multimodal” is one of Pichai’s favorite things about the Gemini AI model—one of the elements that Google claims sets it apart from the guts of OpenAI’s ChatGPT and Microsoft’s Copilot AI assistants, which are also powered by OpenAI technology. It means that Gemini was trained with data in multiple formats—not just text, but also imagery, audio, and code. As a result, the finished modal is fluent in all those modes, too, and can be prompted to respond using text or voice or by snapping and sharing a photo.

“That’s how the human mind works, where you’re constantly seeking things and have a real desire to connect to the world you see,” Pichai enthuses, saying that he has long sought to add that capability to Google’s technology. “That’s why in Google Search we added multi-search, that’s why we did Google Lens [for visual search]. So with Gemini, which is natively multimodal, you can put images into it and then start asking it questions. That glimpse into the future is where it really shines.”

[ad_2]

Source link

February 8, 2024
AI tech is driving the future of education

[ad_1]

Professor Haithem Marzouki, Director of Innovative Pedagogy at NEOMA Business School, discusses how AI technology is transforming educational experiences.

The business education sector has experienced a technological revolution in the last five years. COVID-19 resulted in a sharp rise in online learning, with many business schools investing in developing new programs for remote learners. Developments in generative AI technologies have also provided a host of new tools for professors to use in and outside the classroom.

Implementing these technologies is no easy task. Concerns have been raised that they could impinge on human-led teaching or that students could use them to complete written assignments. Strong communication is an important first step that should precede introducing new learning tools.

At NEOMA Business School, we have clearly seen Artificial Intelligence as a tool to enhance, not replace, human intelligence. In the classroom, AI assists faculty and students, but teaching and assessment methods are being restructured so that they cannot be aced by simply relying on the support of a tool like ChatGPT.

Part of this process involves placing a greater focus on experiential learning. For instance, creating AI-driven business simulations allows students to run virtual companies. They make decisions on a range of issues relating to marketing, production, HR, and finance, and AI tech simulates the consequences of their decisions. The result is an exercise that allows students to learn business strategies in an immersive yet risk-free environment.

In 2023, NEOMA launched a series of new ‘iLearning’ courses designed around this concept. The courses are for remote learners and are structured like a television series. The student is cast in a specific role, such as the Chief Digital Officer at a firm, and gains knowledge through making decisions, completing tasks, and offering analysis or recommendations depending on where their story takes them.

The chief benefit of this approach is that the student cannot be passive; they must interact with the course materials to progress. This is especially important in the context of remote learning, where business schools must be mindful that they are competing with a host of distractions to keep each student’s attention fixed on their learning.

Rise of the cyber teaching assistants

For students who are physically present on campus, AI tech can be used to generate resources. Large Language Model AIs like ChatGPT can instantaneously process, summarise, and evaluate vast quantities of text.

They can be used to provide end-of-class quizzes to check knowledge, offer feedback on students’ written assignments, and even be leveraged as a debate partner during revision activities.

In finance classes, students can use AI tools for financial modelling and forecasting, helping students understand investment strategies, risk management, and financial planning.

In entrepreneurship courses, AI tools aid students in creating robust, data-driven business plans by providing market and competitor analysis. In HR management classes, AI tools can simulate various scenarios, allowing students to experiment with different management strategies.

In addition to generating resources, AI tech can be used to allocate them effectively and customise students’ learning paths by conducting regular performance assessments. Factoring in their learning pace and preferences, the AI can direct resources and faculty attention to students who need more support.

Creating a tailored learning experience

Business schools should consider developing AI-based chatbots or virtual tutors capable of providing students with instant help, explanations, and guidance. These AI assistants could be used to dispense customised advice and resources based on each student’s progress and specific obstacles.

© shutterstock/sdecoret

AI should also be used to enhance automated feedback systems, delivering swift, tailored feedback on students’ work and assessments. The benefit of implementing AI in these systems is that it can be used to identify errors, elucidate them, and offer educational resources to address foundational concepts.

For instance, in 2020, Professor Laura Trinchera at NEOMA transformed her Statistics course into a blended learning format, utilising an AI tool called ALEKS. This AI system, developed by McGraw Hill Education, can quickly and accurately evaluate each student’s performance and provide a fully personalised learning experience.

ALEKS knows if and when a student is ready to progress to a new topic and uses this knowledge to make learning more efficient and effective by continually offering students a selection of topics they are sufficiently prepared to tackle. This builds the student’s learning momentum and confidence.

Embracing the revolutionary potential of AI tech

At the heart of how we introduce AI ‘assistants’ in classes is the principle that AI should help us create a learning experience that is fine-tuned to the individual needs of each student. In this way, we emphasise not what the technology can do but how it can help our students and professors perform to the best of their abilities.

To this end, implementing generative AI tech is not just enhancing educational experiences – it’s fundamentally transforming them.

As business schools adapt to and adopt these digital tools, the focus should remain on cultivating inclusive, stimulating, and dynamic educational environments. The future of education, driven by AI and technology, is not just a concept; it’s a rapidly unfolding reality.

[ad_2]

Source link

February 8, 2024