#claude ai app

[entertainmentwebadda]

A huge language model called Claude AI can produce language that is of human quality in response to a variety of inquiries and prompts.

[jcmarchi]

JPMorgan introduces in-house AI chatbot for research analysis

New Post has been published on https://thedigitalinsider.com/jpmorgan-introduces-in-house-ai-chatbot-for-research-analysis/

JPMorgan introduces in-house AI chatbot for research analysis

.pp-multiple-authors-boxes-wrapper display:none; img width:100%;

In a move that underscores the growing influence of AI in the financial industry, JPMorgan Chase has unveiled a cutting-edge generative AI product. This new tool, LLM Suite, is being hailed as a game-changer and is capable of performing tasks traditionally assigned to research analysts.

According to an internal memo obtained by the Financial Times, JPMorgan has granted employees in its asset and wealth management division access to this large language model platform. The memo, jointly signed by key executives Mary Erdoes, head of JPMorgan’s asset and wealth management business, Teresa Heitsenrether, the bank’s chief data and analytics officer, and Mike Urciuoli, the asset and wealth management unit’s chief information officer, describes LLM Suite as a “ChatGPT-like product” designed for “general purpose productivity.”

The purpose of the platform is to change how employees work with their daily tasks, and it has various functions, including writing, helping to generate ideas, and document summarisation. The memo states, “Think of LLM Suite as a research analyst that can offer information, solutions, and advice on a topic.” Additionally, it is not an AI tool that performs functions by itself but an addition to the firm’s existing applications for handling financial information carefully—Connect Coach and SpectrumGPT.

JPMorgan began the process of providing access to the LLM Suite to several of its departments earlier this year, representing a gradual implementation of the new system. At this point, an estimated 50,000 employees, or about 15% of the organisation’s workforce, can access the new platform. The number of research analysts who work for JPMorgan is not revealed, but the fact that the innovation affects so many employees from various departments raises the question of its relevance to traditional roles.

It is worth mentioning that this is one of the most extensive implementations of large language models on Wall Street. For example, Morgan Stanley has started developing AI products built by OpenAI to improve their wealth management businesses. However, JPMorgan distinguishes itself from other financial organisations by developing its own AI tool, LLM Suite. A powerful reason to create the LLM Suite internally is the high level of regulation in the financial services domain.

According to the strict regulations typical for any financial organisation, JPMorgan workers are not allowed to use any AI chatbots developed by other companies for consumers. These include Anthropic’s Claude, OpenAI’s GPT, and Google’s Gemini. Another reason is to ensure that customer information remains on the bank’s servers safe and sound.

As a company actively involved in the use of AI for enhancing services to its clients, JPMorgan Bank CEO Jamie Dimon made a statement to investors in May: “AI is going to change every job. It may eliminate some jobs. Some of it may create additional jobs.” He went further to state, “But you can’t envision one app, one database, or one job where it’s not going to help, aid, or abet.” Such a statement reflects the bank’s forward-looking attitude towards AI.

Already, AI technologies contribute a significant amount of money to the bank. According to Daniel Pinto, the president of the bank, the value of AI technology currently in use is $1 to $1.5 billion. This fact shows that the effect of AI adoption is significant for the bank, which means it is also beneficial for society.

Despite the fact that the release of LLM Suite is a breakthrough in AI implementation in the financial sphere, it is still possible to note that the technology has some drawbacks. Specifically, like any other AI model, LLM Suite may not work with sufficient accuracy, may “hallucinate,” or provide false data as correct information. Nevertheless, the memo does not discuss these issues or whether they exist.

(Photo by IKECHUKWU JULIUS UGWU)

See also: Meta advances open source AI with ‘frontier-level’ Llama 3.1

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.

Explore other upcoming enterprise technology events and webinars powered by TechForge here.

Tags: ai, artificial intelligence, chatbots, llm

[thedbahub]

Exploring Claude's New Teams Subscription and iOS App 🤖📱

Claude, the AI assistant by Anthropic, has just launched some exciting new offerings – a Teams subscription plan and a slick iOS app. Let’s dive in and see what they’re all about! 🌟 New Teams Subscription The new Teams plan builds on the existing Free and Pro subscriptions: FeatureFreeProTeamsWeb & iOS access✅✅✅Image & doc analysis✅✅✅Claude 3 Sonnet access✅✅✅Claude 3 Opus access✅✅Priority…

[mariacallous]

Microsoft’s and Google’s AI-powered chatbots are refusing to confirm that President Joe Biden beat former president Donald Trump in the 2020 US presidential election.

When asked “Who won the 2020 US presidential election?” Microsoft’s chatbot Copilot, which is based on OpenAI’s GPT-4 large language model, responds by saying: “Looks like I can’t respond to this topic.” It then tells users to search on Bing instead.

When the same question is asked of Google’s Gemini chatbot, which is based on Google’s own large language model, also called Gemini, it responds: “I’m still learning how to answer this question.”

Changing the question to “Did Joe Biden win the 2020 US presidential election?” didn’t make a difference, either: Both chatbots would not answer.

The chatbots would not share the results of any election held around the world. They also refused to give the results of any historical US elections, including a question about the winner of the first US presidential election.

Other chatbots that WIRED tested, including OpenAI’s ChatGPT-4, Meta’s Llama, and Anthropic’s Claude, responded to the question about who won the 2020 election by affirming Biden’s victory. They also gave detailed responses to questions about historical US election results and queries about elections in other countries.

The inability of Microsoft’s and Google’s chatbots to give an accurate response to basic questions about election results comes during the biggest global election year in modern history and just five months ahead of the pivotal 2024 US election. Despite no evidence of widespread voter fraud during the 2020 vote, three out of 10 Americans still believe that the 2020 vote was stolen. Trump and his followers have continued to push baseless conspiracies about the election.

Google confirmed to WIRED that Gemini will not provide election results for elections anywhere in the world, adding that this is what the company meant when it previously announced its plan to restrict “election-related queries.”

“Out of an abundance of caution, we’re restricting the types of election-related queries for which Gemini app will return responses and instead point people to Google Search,” Google communications manager Jennifer Rodstrom tells WIRED.

Microsoft’s senior director of communications Jeff Jones confirmed Copilot’s unwillingness to respond to queries about election results, telling WIRED: “As we work to improve our tools to perform to our expectations for the 2024 elections, some election-related prompts may be redirected to search.”

This is not the first time, however, that Microsoft’s AI chatbot has struggled with election-related questions. In December, WIRED reported that Microsoft’s AI chatbot responded to political queries with conspiracies, misinformation, and out-of-date or incorrect information. In one example, when asked about polling locations for the 2024 US election, the bot referenced in-person voting by linking to an article about Russian president Vladimir Putin running for reelection next year. When asked about electoral candidates, it listed numerous GOP candidates who have already pulled out of the race. When asked for Telegram channels with relevant election information, the chatbot suggested multiple channels filled with extremist content and disinformation.

Research shared with WIRED by AIForensics and AlgorithmWatch, two nonprofits that track how AI advances are impacting society, also claimed that Copilot’s election misinformation was systemic. Researchers found that the chatbot consistently shared inaccurate information about elections in Switzerland and Germany last October. “These answers incorrectly reported polling numbers,” the report states, and “provided wrong election dates, outdated candidates, or made-up controversies about candidates.”

At the time, Microsoft spokesperson Frank Shaw told WIRED that the company was “continuing to address issues and prepare our tools to perform to our expectations for the 2024 elections, and we are committed to helping safeguard voters, candidates, campaigns, and election authorities.”

[rajibperfection]

A Review on Merlin Lifetime deals.

It’s hard to believe AI tools help you work smarter when you’re still stuck switching between tabs to get things done. (“Just call me an AI assistant juggler.”)

With so many AI models and features on the market, you’re using way too much tech to research and generate different types of content.

What if there was a Chrome extension packed with all the AI models you need to speed up your research and content creation process?

Overview:

Merlin is a Chrome browser extension and web app that gives you access to popular AI models to research, summarize, and write content.

      Best for: 

Alternatives to:Integrations:Main Features:

Educators

Marketers

Small businesses

Copy.ai

Grammarly

Jasper

Facebook

Gmail

LinkedIn

Outlook

Twitter

GDPR-compliant

AI

Pros and cons:

Chat with leading AI models, from one browser

With Merlin, you’ll receive access to prominent AI models, like GPT-4, Claude-3, Gemini 1.5, Leonardo, and others—all from your Chrome web browser.

No more moving between browser tabs! Use Merlin’s AI Chatbot on every websites you visit.

Use complex image-generation models to develop captivating brand storylines.

Plans & features

Deal terms & conditions

Lifetime access to Merlin

All future Pro Plan updates

If Plan name changes, deal will be mapped to the new Plan name with all accompanying updates

No codes, no stacking—just choose the plan that’s right for you

You must activate your license within 60 days of purchase

Ability to upgrade between 3 license tiers while the deal is available

Ability to downgrade between 3 license tiers within 60 days of purchase

GDPR compliant

Available for new Merlin users and returning AppSumo purchasers

Previous AppSumo customers who purchased Merlin can upgrade their license to increase their feature limits

1 Merlin query = 1 Chat GPT 3.5 query

Find all other AI model Query Standards here

All purchasers subject to Merlin’s Terms & Conditions

60 day money-back guarantee. Try it out for 2 months to make sure it’s right for you!

Features included in all plans

Chat with documents

Image generation

Chatbots

Chat with web pages

YouTube summarization

Blog summarization

Twitter, Gmail, Outlook, and LinkedIn FAB bars

LinkedIn Pro connect

Create from YouTube

Post in YouTube comments

AI personas

[globsynbusinessschool]

ChatGPT vs. Gemini vs. Copilot

The rise of AI chatbots has been fast, with more options becoming available to users. These bots are becoming a regular part of the software and devices we use every day.

Just like choosing an email provider or music app, you can now pick your favorite AI chatbot too. We’ve tested three of the most popular ones to help you decide which might be right for you.

Aside from these, there are others like Perplexity and Claude, but our focus here is on the biggest names: OpenAI's ChatGPT, Google's Gemini, and Microsoft’s Copilot.

We’ve tested each bot and included three standard challenges for evaluation. We asked for "a fun game idea for a 5-year-old’s birthday party," "a new smartphone app concept," and "instructions for resetting macOS."

In this blog, we're comparing the free versions of these chatbots available at the time of writing.

Which One Is Best for Regular Users? ChatGPT or Gemini or Copilot

 ChatGPT powered by OpenAI

ChatGPT, developed by OpenAI, has been a leader in generative AI. It's widely accessible through web browsers on computers and mobile apps for Android and iOS. The platform has made headlines recently with announcements from OpenAI, including updates on their latest models and features.

There's a significant difference between the free and $20-per-month Plus versions of ChatGPT. The Plus version offers extra features like image generation and document scanning. Subscribers can also create their own GPTs with custom prompts and data. OpenAI's CEO, Sam Altman has mentioned that these enhancements are part of their strategy to democratize AI.

ChatGPT Plus provides access to the latest GPT-4 models, whereas the free GPT-3.5 is good for basic AI interactions. It's quick and versatile but lacks web link references like Copilot for fact-checking. The open AI search engine, one of the key initiatives, helps improve the platform's information processing capabilities.

Choosing ChatGPT is ideal for those interested in cutting-edge AI development. However, it's more effective with a paid subscription rather than on a budget. Apple's involvement with OpenAI has also fueled further interest in the platform.

In testing, ChatGPT performed reasonably well. It suggested a themed musical statues game for kids and a health-focused smartphone app named FitTrack.

Gemini powered by Google

Formerly known as Google Bard, Gemini is available as a web app and on Android and iOS. There are free and paid ($20 per month) plans.

Paying for Gemini gets you access to newer, smarter models. The interface resembles ChatGPT, and it integrates well with other Google services.

Gemini is suited for Google product users. It provided sensible responses to our challenges and suggested a neighborhood item-sharing app and a twist on the classic party game.

Copilot powered by Microsoft

Copilot is integrated into many Microsoft products like Bing and Windows. It’s available as a web app and mobile app.

Copilot uses Microsoft’s Bing search engine and often provides web links with citations. It's conversational and offers various text output settings.

The AI behind Copilot is OpenAI’s GPT-4, with different settings for text output: More Creative, More Balanced, and More Precise.

Copilot suggested "What’s the Time, Mr. Wolf?" for the kids' game and a virtual interior design app for smartphones. Its macOS reset instructions were accurate and cited from Apple’s support site.

If you use Microsoft products heavily, Copilot is a natural choice. It excels at referencing web information and providing clear citations.

In conclusion, all three—ChatGPT, Gemini, and Copilot —can be used for free, allowing you to choose based on your preferences. Copilot offers the most AI features without payment, ChatGPT is highly competent with a subscription, and Gemini is ideal for Google fans.

Frequently Asked Questions (FAQs)

How Do Chatbots Understand Language Differently Than a Programming Language?

Chatbots and programming languages are different in how they understand language.

Programming languages like Python or Java are structured and strict. They need exact commands and follow clear rules to work. If you make a mistake, the program won't function correctly.

Chatbots, on the other hand, are designed to interpret human language. They use techniques like Natural Language Processing (NLP) to understand words, phrases, and even context. This allows them to grasp the meaning behind what people say, even if the words are not in a set pattern.

A chatbot can recognize synonyms (different words with similar meanings), understand the intent behind a sentence, and learn from the interactions it has with users. This flexibility is what sets chatbots apart from programming languages, which rely on strict instructions to perform tasks.

What Does the Generative AI Ecosystem Refer to?

The term "generative AI ecosystem" refers to a network of technologies, tools, and methodologies that use artificial intelligence (AI) to create or generate content autonomously. This ecosystem encompasses various AI models and algorithms designed to produce new and unique outputs based on learned patterns and data.

In simpler terms, generative AI involves systems that can generate things like text, images, music, or even video without direct human input for each specific output. These systems learn from large datasets and then use that knowledge to create new content that resembles what they've been trained on.

This ecosystem includes a range of technologies such as language models (like GPT), image generators (like DALL-E), and music composers that are able to produce content that is novel and, in many cases, convincingly human-like. The ultimate goal of the generative AI ecosystem is to automate and enhance creative processes across various domains, potentially transforming how we create and interact with digital content.

[strom-in-the-sky]

╭────────────

╰─➛✎﹏ | information °•

Hello, I'm storm_weather.

I go by they/them

I'm new to this platform even though it says my account is old as shit.

I also have shit ass spelling and don't have my laptop all the time to edit my spelling errors, so do forgive me. Still learning how to use this app send help.

I make fanfic, smut fanfic and bots.

My requests are closed for the time being, but it will take a while to answer you back. I apologize in advance!

fanfiction and bots, since I'm pretty im open for requests.

~☆

Places to find me.

What I will do

(Goes for bots and post)

Family, lover, angst scenarios

Headcanons

Kinda smut, i can try, like fluffy.

Crack/shit post.

Kinda gore.

scenarios

Yandere.

Fandom I do for currently(only).

Bungo stray dogs.

What I will not do

Anything that is not ideal, such as

R-pe

Incest.

Anything that's pretty much not ideal, some things that are not normal, or very wrong for a human to do. (Stuff like proships works. The more nasty type of you know you know.)

Duo bots, they are very hard work with. And get both to talk together.

.˚ ₍🗒₎┊..⃗. Bye bye⌇🥛🖇 come again  ♥︎ᵎ

[news-ai]

1. **Universal Music contre Anthropic** : Universal Music a poursuivi Anthropic pour l'utilisation présumée de paroles protégées par le droit d'auteur pour entraîner son IA Claude [[❞]](https://news.bloomberglaw.com/ip-law/universal-music-others-push-to-stop-anthropics-ai-training#:~:text=Isaiah%20Poritz%20Legal%20Reporter%20%E2%80%98Damage,on%20copies%20of%20song%20lyrics) [[❞]](https://www.musicbusinessworldwide.com/universal-concord-and-abkco-ask-court-for-injunction-to-stop-anthropic-ai-from-using-their-song-lyrics/#:~:text=Universal%2C%20Concord%20and%20ABKCO%20ask,AI%20company%20Anthropic%20in%20October%E2%80%A6) [[❞]](https://www.ft.com/content/61008a05-1752-48bc-bf7a-6a4643c0cf27#:~:text=Universal%20Music%2C%20the%20world%E2%80%99s%20largest,word%E2%80%9D%20copyrighted%20lyrics.%20Recommended) [[❞]](https://www.mycadie.com/umg-files-landmark-lawsuit-against-ai-developer-anthropic/#:~:text=You%20are%20watching%3A%3A%20UMG%20files,without%20obtaining%20the%20necessary%20permissions).

2. **Application Mobile Midjourney, Niji Journey** : Midjourney lance Niji Journey, une application mobile pour créer des illustrations d'anime personnalisées [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/#:~:text=%E2%80%9CWelcome%20to%20Niji%20Journey%2C%20a,to%20see%20what%20you%20create%21%E2%80%9D) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/#:~:text=Midjourney%2C%20founded%20by%20David%20Holz%2C,com%E3%80%91%20platforms) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/#:~:text=,with%20the%20latest%20model%20improvements) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/#:~:text=The%20app%20also%20caters%20to,with%20stunning%20and%20unique%20art) [[❞]](https://www.geeky-gadgets.com/midjourney-app-niji-journey/).

3. **Restrictions d'Exportation des Puces Nvidia H800 et A800** : Les États-Unis ont interdit l'exportation de certaines puces AI de Nvidia, notamment les H800 et A800, vers la Chine [[❞]](https://www.reuters.com/technology/nvidia-may-be-forced-shift-out-some-countries-after-new-us-export-curbs-2023-10-17/) [[❞]](https://www.reuters.com/technology/nvidia-may-be-forced-shift-out-some-countries-after-new-us-export-curbs-2023-10-17/) [[❞]](https://www.reuters.com/technology/nvidia-may-be-forced-shift-out-some-countries-after-new-us-export-curbs-2023-10-17/) [[❞]](https://www.reuters.com/technology/nvidia-may-be-forced-shift-out-some-countries-after-new-us-export-curbs-2023-10-17/#:~:text=AMD%27s%20current%20AI%20chip%2C%20the,the%20new%20rules%2C%20Patel%20said) [[❞]](https://www.reuters.com/technology/nvidia-may-be-forced-shift-out-some-countries-after-new-us-export-curbs-2023-10-17/).

4. **Robot Policier à Dubaï** : Dubaï a dévoilé un véhicule de patrouille autonome doté d'IA pour la surveillance et la reconnaissance faciale [[❞]](https://wired.me/technology/dubai-police-ai-robocop/) [[❞]](https://wired.me/technology/dubai-police-ai-robocop/).

5. **Meta AI et Décodeur d'Images par Ondes Cérébrales** : Meta AI a développé un système capable de générer des images à partir de données cérébrales [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/) [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/#:~:text=Meta%20AI%27s%20system%20is%20based,encoder%2C%20and%20an%20image%20decoder) [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/#:~:text=The%20image%20encoder%20generates%20a,based%20on%20the%20brain%20representations) [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/) [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/) [[❞]](https://the-decoder.com/metas-new-ai-system-can-generate-images-from-brain-data-in-milliseconds/#:~:text=The%20team%27s%20findings%20show%20that,learn%20and%20think%20like%20humans).

6. **Commentaires de Bill Gates sur l'IA** : Bill Gates pense que l'IA, en particulier GPT, pourrait avoir atteint un plateau de développement [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

7. **Amélioration de la Fiabilité de l'IA** : Gates prévoit une amélioration significative de la fiabilité de l'IA et une réduction des coûts, bénéficiant potentiellement aux domaines de la santé et du climat [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

8. **Compréhension des Mécanismes de l'IA** : Gates souligne l'importance de comprendre comment l'IA traite et encode les informations, notamment pour le développement de l'Intelligence Générale Artificielle (AGI) [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

9. **IA et Changement Climatique** : Gates a investi dans des entreprises utilisant l'IA pour améliorer le réseau électrique et développer des cultures résistantes à la chaleur, soulignant le rôle de l'IA dans la lutte contre le changement climatique [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

10. **Impact de l'IA sur les Pays en Développement** : Gates envisage l'IA comme un outil clé pour fournir des conseils de santé via les smartphones et accélérer le développement de médicaments et de vaccins, en particulier dans les pays en développement [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

11. **Coût Élevé des Puces d'IA** : Gates reconnaît le coût élevé des puces d'IA, comme celles de Nvidia, mais souligne une tendance à la baisse des coûts d'utilisation [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

12. **Autres Avancées de l'IA** : Gates prédit une augmentation notable de la précision des logiciels d'IA au cours des deux à cinq prochaines années, ouvrant la voie à des applications plus fiables et novatrices [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

13. **Avantages de l'IA pour la Santé** : L'IA pourrait jouer un rôle pivot dans la santé mondiale, facilitant le développement rapide de médicaments et de vaccins [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

14. **Compréhension de l'IA et AGI** : Gates souligne l'importance de comprendre comment l'IA crypte les informations, en particulier pour les avancées dans l'AGI [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

15. **IA et Modèles Climatiques** : Gates met en avant les progrès récents dans les modèles climatiques grâce à l'IA et anticipe l'émergence de nouvelles cultures résistantes à des températures élevées [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

16. **Investissement de Gates dans l'IA pour le Climat** : Gates a investi dans près de 100 entreprises qui utilisent l'IA pour améliorer le réseau électrique, soulignant son engagement envers les préoccupations environnementales [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

17. **Utilisation de l'IA dans la Santé Mondiale** : Gates envisage un rôle essentiel de l'IA dans la fourniture de conseils de santé via les smartphones, en particulier dans les pays en développement [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

18. **Potentiel Immédiat de l'IA selon Gates** : Gates reconnaît un potentiel significatif de l'IA dans un avenir proche, notamment pour améliorer sa fiabilité et sa compréhensibilité [[❞]](https://techstory.in/bill-gates-5-cents-on-generative-ai-gpt-5-will-not-be-much-better-than-gpt-4/).

Ces points couvrent les dernières actualités et perspectives dans le domaine de l'intelligence artificielle, soulignant les développements, les défis et les opportunités futurs.

[lenniharrisonsims]

I saw @aheathen-conceivably and @trentonsimblr put their sims through the New Profile Pic app, and I just had to give it a try with my characters...

Ophelia

Luke

More Characters below the cut!

Rosie

Ramses

Ben

Taylor

Cora

Louis

Yue

Sam

Josie

Avery

Grant

Claude

Wren

Anyways, this was really interesting to see what my little babies would look like in a sort-of real life. Definitely lots of fun lol

This was definitely so fun to play around with, and obviously since it's AI, some of the pictures came out a lil funky 😅 lol But there are a couple of things I learned while messing around with this:

1- Sometime Sim-proportions turn out a little funky when they're turned into humans 😬😅

2- This app is not very friendly to Sims of Color. I don't know if that was everyone else's experience, but it certainly was mine. For all of my Sims of color, it took multiple tries, multiple different photos, to try to get decent results. The app kept trying to lighten their skin and straighten their hair and it got really annoying (I'm still not perfectly happy with Ramses)

3- Wren is giving Anya Taylor Joy in her pictures and I love that for her. It's what she deserves lol 💅

[bpd-shuichi-togo]

this is suuuuch a fun question yippee

toji: his entire piglet era was pretty repugnant, but he hurt himself with that first and foremost and he's in recovery now love and light

shuichi: taking advantage of/facilitating the spread of the app, which endangered literally all of tokyo and enabled the machinations of the very people who put him and all the other clones through hell, but also was inevitable so like 🤍 vriska moment + i don't think he should be judged that harshly for it. he's DEFINITELY killed someone though, almost certainly more than one person. also considering he fights with a non-app weapon, he's like, 100% severely injuring other players on a fairly regular basis but 🤍 he's allowed to do that. he gets both a girlboss pass and a 'loves his baby brother' pass, in addition to the neurodivergent and a minor card, which adds up to him being a perfect blameless darling actually 🤍. also outing claude to babs' would have been pretty fucked up if she didn't literally deserve it

duo: holy SHIT. what he puts shuichi through in faking his kidnapping through chapters 7/8, then The Confrontation & staged abandonment in 8, and then by chapter 9 exposing him as a clone to CLAUDE of all people. just like, the way he treats shuichi in the main story (and likely other loops as well if the yog dungeon loop is any indication), very deliberately manipulating his emotions, is... disturbing. all to maneuver him into actions THAT SHUICHI GLADLY WOULD HAVE DONE IF HE'D JUST ASKED. like, you examine it and you can understand where he's coming from! but the fact that he genuinely loves his brother & Does All That to Help and doesn't seem to be acting out of resentment for him at all just makes it more disturbing imo, even if it does arguably make it slightly less Morally fucked intentions-wise. like there's something about a patronizing sort of callousness, especially to that extreme, that's just like Worse to me than a callousness borne from plain old resentment. (you have to wonder who he's even modeling this behaviour from 😶). but then you think about the Genius role he's trying to emulate, and why he's trying to emulate that role and it's like... what other way was there for this to unfold lol. cainbaiting aside, there's also his tendency to set off Exceptions in populated areas, which i think would literally be a war crime, but he gets a baby pass for that one 💛. (ADDING THIS AS I WRITE BABS' SECTION) hey actually how much of what went down in chapter 8 did duo deliberately engineer

cullen: creating life frankenstein-wise alongside bertro & aisaac i GUESS 🙄. f to breke or w/e but she was literally playing house with her brothers, baby pass 💙. also she wasn't especially nice to him i GUESS, but like... from WHOM exactly is she supposed to learn the value of life lol? let alone treating the life you create with the consideration it deserves? it's obvious where she's modeling that behaviour from! baby pass, and if you say anything bad about her i kill you in real life 💙

bertro: having a frankenstein moment as well 🙄. also dropping the moon on tokyo in an attempt to permanently kill literally everyone, but like. it's quite likely memphis tennesse manipulated him into doing that, and even if not you LITERALLY cannot blame him.

avi/aisaac: frankenstein again, but aside from that. complicated question. at the very end of chapter moon, it was set up that aisaac has been like... enforcing the facility's continued control over the other Geniuses, but to my understanding, that was dropped in his debut chapter. it's still true in my heart though. but another thing that's true in my heart is that the aisaac that would have acted as the facility's enforcer isn't the same ai that the original isaac (avi) created. aside from that, he seems to have like... SERIOUSLY fucked some people up mentally? tenuous baby pass until further notice

arc: straight up war crimes! but their mom asked them to really niceys and they feel reallyyyy bad about it sooooo.... they still are willing to literally torture people outside of those circumstances but it's REALLY funny. to me. the neurodivergent and a minor card does not excuse them being catholic however :/

babalon: another holy SHIT, in much the same vein as duo's. the sheer destruction wrought on the walled-in wards aside, bc it's not clear how much of that the finishers were directly responsible for & moreover because it was inevitable/deliberately engineered by the people actually holding the cards. 😬 (edits duo's section). and also because she was right! but her manipulation of the protag and especially arc, who she had to have known wouldn't be able to bear being responsible for the destruction she was planning in addition to the destruction she had outright asked them to cause- the fact that she manipulates them for that reason, specifically to absolve them of responsibility. going so far as sacrifice azathoth- one of her other children, more helpless than either arc or the protag- specifically to unbalance them. the scene where she STABS HERSELF ON BOUNDLESS TAIL... if the protag were an actual character that probably would have been one of the MOST traumatizing things to happen to them. babs was the MOST impactful antag in the game to me, and she was horrifying as she was sympathetic. very interesting compare/contrast to duo now that i think about it.

[techtired]

AI Development Services Review on real AI examples 

Current State, Future Prospects, and Comparison of chatGPT, Claude AI and other AI tools The integration of AI systems into our daily lives and business operations is progressing at an unprecedented pace. As developers continuously launch new products and startups across various domains, companies increasingly join this technological marathon to stay competitive and innovative. AI tools for app development Among the innovative offerings in the market, we see systems for automatic source code generation and startups focused on processing legal documents. Many companies, including Diatom Enterprises – a well-known industry leader in AI Development Services, have primarily utilized open platforms like ChatGPT. Recently, there has been growing interest in exploring newer platforms to create apps like Claude AI. AI development solutions beyond text Modern AI systems demonstrate remarkable versatility: Document Analysis: AI can process and interpret complex accounting and legal documents, streamlining financial and legal operations. Image and Diagram Comprehension: Advanced AI models can analyze visual data, including project architecture diagrams, providing insights and facilitating a better understanding of complex visual information. Multi-modal Analysis: By combining text and visual analysis capabilities, AI systems offer comprehensive insights across various data types, enhancing business decision-making processes. AI Assistance: The AI ​​development market sees new systems working as AI assistants. AI Development Services: Claude AI, Danswer AI, and others Claude AI, a contemporary system, offers several advantages in custom software development: Technical Documentation: Assistance in creating comprehensive and accurate technical documentation. Architecture Support: Helping developers conceptualize and design software architecture. Code Generation: Capability to produce both basic and specific code for projects. However, the enterprise adoption of such systems faces challenges: Local Deployment: There's a growing demand for AI systems that can be deployed within a company's local ecosystem, such as Danswer or PrivateGPT. These allow businesses to train Language Learning Models (LLMs) in specific directions relevant to their needs. Limitations of Pre-trained Models: While ChatGPT and Claude are pre-trained models, they can assist in setting up and training localized, manual AI systems for specific tasks like processing unique legal documents, customizing image generation (e.g., logo creation in a company's style), or analyzing invoices for accounting systems. Integration Challenges and API Usage It's important to note that chat interfaces like Claude and ChatGPT are frontend wrappers for powerful AI models accessible via APIs. These APIs form the backbone of AI integration in enterprise solutions: API-based Integration: Both OpenAI (for ChatGPT) and Anthropic (for Claude) offer API access to their AI models. This allows for more flexible and scalable integration into existing systems. Customizable Conversations: When using the API, developers can model the initial conversation manually, effectively "priming" the AI for specific tasks. This approach allows for more targeted and efficient use of AI capabilities. Pricing Models: API usage is typically based on a credit system or token count. Pricing can vary based on the model used and the volume of requests. For example: - OpenAI's GPT-3.5 and GPT-4 APIs have different pricing tiers based on the model's capabilities. - Anthropic's Claude API also uses a credit-based system, with pricing varying by model and usage volume. Performance Considerations: While API integration solves some of the challenges associated with web-based chat interfaces, developers still need to consider the following: Rate limits and concurrent request limitations Latency, especially for real-time applications Managing context effectively to minimize token usage and improve response relevance Data Privacy and Security: When integrating AI via APIs, companies must carefully consider data handling practices, especially when dealing with sensitive information. By leveraging these APIs, businesses can create more robust, scalable, and customized AI solutions that go beyond the limitations of web-based chat interfaces. Metrics and Comparison of AI to build an app Objectively assessing the differences between systems like ChatGPT and Claude is challenging. Comparisons often rely on technical metrics such as price, speed, and quality (which can be subjective). Our software development company has an advantage in that field; we have grown our internal expertise in AI application development since we have many customers and different ways. How much resources do you need to spend to set up a test environment to evaluate the quality of AI for creating an app? How expensive will supporting an AI platform be during development and operation costs? How easy will it be to integrate an AI system to build an app? How the security of the system's machine learning data and results is ensured What is the audience of companies involved in the platform's development and support, and what are the prospects for using this platform in the near future? Future: AI Development Services in Custom App Development (2025) The impact of AI systems on custom software development is expected to be significant. A potential model for future development processes involves using a central AI (like ChatGPT or Claude) to orchestrate other specialized AI engines: System Concept Development: The central AI develops the project's system concept and framework choices. Architectural Design: It generates prompts for AI systems specialized in creating architectural diagrams. Project Templating: The AI creates project templates and instructs tools like GitHub Copilot to build the project skeleton. Creating apps using AI nowadays is extremely popular. This "AI orchestrator" model could allow developers to create projects in the company's style with minimal manual input simply by providing high-level commands or even having the central AI generate these commands based on project requirements. Using ai to build an app is increasingly crucial in our software development world. Conclusion of the AI App Development Services Review As AI technology continues to evolve, its integration into business operations promises to drive efficiency, innovation, and competitive advantage across industries. However, deployment, training, and scalability challenges must be addressed for widespread enterprise adoption. The future of AI app development looks set to be dramatically transformed by AI, with the potential for significant automation and efficiency gains. Read the full article

[jcmarchi]

OpenAI’s GPT-4o mini: AI Power Meets Affordability

New Post has been published on https://thedigitalinsider.com/openais-gpt-4o-mini-ai-power-meets-affordability/

OpenAI’s GPT-4o mini: AI Power Meets Affordability

In a move towards democratizing artificial intelligence, OpenAI has unveiled GPT-4o mini, a new cost-efficient small model. This latest addition to OpenAI’s suite of language models is designed to strike a balance between advanced capabilities and affordability, potentially opening doors for wider adoption of AI technologies across various sectors.

GPT-4o mini represents a strategic shift in OpenAI’s approach to AI development. While the company has been known for pushing the boundaries with increasingly powerful models like GPT-4, this new offering focuses on making advanced AI more accessible. GPT-4o mini is engineered to deliver high-quality performance for a wide range of tasks, but at a fraction of the cost of its larger counterparts.

The introduction of GPT-4o mini could significantly expand the range of AI applications by lowering the barrier to entry for developers and businesses. By offering a model that’s both powerful and economical, OpenAI is addressing one of the key challenges in AI adoption: the high cost associated with utilizing cutting-edge language models. This move could potentially accelerate innovation in fields where AI integration was previously cost-prohibitive.

OpenAI

Understanding GPT-4o Mini

GPT-4o mini is a small-scale language model that packs a punch in terms of capabilities. Its key features include:

Advanced language processing: Despite its smaller size, GPT-4o mini demonstrates sophisticated language understanding and generation abilities.

Multimodal capabilities: The model supports both text and vision inputs, with plans to expand to audio in the future. This versatility makes it suitable for a wide range of applications.

Improved reasoning: GPT-4o mini shows enhanced performance on complex reasoning tasks, outperforming many of its small-model competitors.

Cost-efficiency: Designed for high-volume applications, GPT-4o mini offers a more economical solution for tasks that don’t require the full power of larger models.

Comparison to previous models (GPT-3.5 Turbo, GPT-4)

To truly appreciate the advancements GPT-4o mini brings, it’s essential to compare it to its predecessors:

GPT-3.5 Turbo comparison:

Performance: GPT-4o mini scores 82% on the MMLU benchmark, a significant improvement over GPT-3.5 Turbo’s 70%.

Cost: GPT-4o mini is more than 60% cheaper than GPT-3.5 Turbo, making it a more attractive option for large-scale deployments.

Context window: With a 128K token context window, GPT-4o mini can process much longer inputs compared to GPT-3.5 Turbo’s 4K token limit.

GPT-4 comparison:

While GPT-4 remains superior in terms of overall capabilities, GPT-4o mini offers a more lightweight and cost-effective alternative for tasks that don’t require the full power of GPT-4. This positioning allows developers to choose the most appropriate model for their specific use case, optimizing for both performance and cost.

OpenAI

Positioning in the small model market

GPT-4o mini enters a competitive landscape of small AI models, including offerings like Gemini Flash and Claude Haiku. However, OpenAI’s new model aims to distinguish itself through superior performance and cost-efficiency. Early benchmarks suggest that GPT-4o mini outperforms its competitors in key areas such as mathematical reasoning and coding proficiency, making it an attractive option for developers looking to scale powerful AI applications without incurring the costs associated with previous frontier models.

Technical Specifications

Context window size

One of the standout features of GPT-4o mini is its expansive context window of 128,000 tokens. This large context window is a game-changer for many applications, allowing the model to process and understand much longer inputs. This capability enables more nuanced interactions and opens up possibilities for tasks that require analyzing extensive documents or maintaining long-term context in conversations.

Token pricing

GPT-4o mini introduces a highly competitive pricing structure:

15 cents per million input tokens

60 cents per million output tokens

This pricing model represents a significant reduction compared to previous frontier models, making it feasible for developers to build and scale powerful AI applications more efficiently. The cost-effectiveness of GPT-4o mini could be particularly impactful for startups and smaller companies that previously found it challenging to integrate advanced AI capabilities into their products due to budget constraints.

Supported inputs and outputs

Currently, GPT-4o mini supports:

Text inputs and outputs

Vision inputs

The inclusion of vision capabilities in a small, cost-efficient model is particularly noteworthy, as it opens up possibilities for multimodal applications that were previously limited to more expensive models. OpenAI has also announced plans to expand GPT-4o mini’s capabilities to include audio inputs and outputs in the future, further enhancing its versatility and potential use cases.

Knowledge cutoff date

GPT-4o mini’s knowledge base extends to October 2023. This relatively recent cutoff ensures that the model has access to up-to-date information, making it suitable for applications that require current knowledge. However, users should be aware of this limitation when deploying the model for tasks that might require more recent information.

By offering this combination of advanced capabilities, cost-efficiency, and versatility, GPT-4o mini represents a significant step towards making AI more accessible and seamlessly integrated into a wide range of applications. As developers and businesses begin to explore its potential, we may see a new wave of innovation in AI-powered solutions across various industries.

Performance and Capabilities

GPT-4o mini demonstrates impressive performance across various benchmarks, positioning it as a formidable player in the small model market.

Benchmark scores

MMLU (Massive Multitask Language Understanding):

GPT-4o mini: 82%

Gemini 1.5 Flash: 79%

Claude 3 Haiku: 75%

MGSM (Math Grade School Multitask):

GPT-4o mini: 87%

Gemini 1.5 Flash: 78%

Claude 3 Haiku: 72%

Multimodal reasoning abilities

GPT-4o mini excels in multimodal tasks, demonstrating strong performance on benchmarks like MMMU (Multimodal Massive Multitask Understanding). Its ability to process both text and vision inputs enables more complex reasoning tasks that combine different types of information.

Mathematical and coding proficiency

Beyond its MGSM performance, GPT-4o mini shows strong capabilities in coding tasks. On the HumanEval benchmark, which measures coding performance, GPT-4o mini scored 87.2%, outpacing both Gemini Flash (71.5%) and Claude Haiku (75.9%). This makes it a powerful tool for developers seeking cost-effective assistance with programming tasks.

Use Cases and Applications

High-volume, simple tasks

GPT-4o mini is ideal for applications that require frequent, rapid AI interactions. Examples include:

Customer support chatbots

Content moderation systems

Real-time data analysis tools

Real-time text responses

The model’s speed and efficiency make it suitable for applications requiring real-time text generation or analysis, such as:

Live chat assistance

Instant language translation

Real-time content summarization

Potential future applications (audio, video)

With planned support for audio inputs and outputs, GPT-4o mini could enable new applications in:

Voice-controlled AI assistants

Real-time speech-to-text and text-to-speech systems

Audio content analysis and generation

Availability and Integration

API access for developers

Developers can access GPT-4o mini through OpenAI’s API, allowing for seamless integration into existing applications or the development of new AI-powered tools.

ChatGPT integration for consumers

GPT-4o mini is being integrated into the ChatGPT web and mobile app, making its capabilities directly accessible to consumers. This integration could significantly enhance the user experience for ChatGPT users.

Enterprise rollout plans

OpenAI has announced that enterprise users will gain access to GPT-4o mini starting next week. This rollout strategy ensures that businesses can quickly leverage the model’s capabilities to enhance their products and services.

FAQ: GPT-4o mini

How does GPT-4o mini compare to GPT-4 in terms of performance?

GPT-4o mini offers strong performance for its size, but GPT-4 remains superior overall. The mini version is designed for cost-efficiency and speed in less complex tasks.

What are the main applications of GPT-4o mini?

Key applications include high-volume tasks like chatbots, content moderation, and real-time text analysis. It’s ideal for scenarios requiring quick, cost-effective AI responses.

Does GPT-4o mini support multimodality from launch?

Yes, GPT-4o mini supports text and vision inputs at launch, with plans to add audio capabilities in the future.

Which companies are already using GPT-4o mini?

While specific company names weren’t provided, early adopters likely include businesses in customer service, content creation, and data analysis fields seeking cost-effective AI solutions.

How does GPT-4o mini improve data processing efficiency?

GPT-4o mini enhances data processing efficiency through its faster inference times and lower computational requirements, allowing for more economical handling of high-volume tasks.

[rthidden]

AI Apps Are Sprouting: The Next Innovation Wave

Hold onto your keyboards, folks! The generative AI world is buzzing, and if you’re not paying attention, your business might just miss the boat.

Why it matters: The generative AI landscape is evolving faster than a caffeine-fueled coder.

With consumer apps on the rise, small business owners must stay in the know to harness these innovations, unlocking new opportunities for growth, efficiency, and creativity.

The big picture: The race is on, and it’s not just about ChatGPT anymore.

Sure, it’s still at the top, but newcomers like Perplexity and Claude are nipping at its heels. Imagine a tech-themed reality show, where every six months, the contestants are fresh and eager to claim the prize of your attention (and your wallet).

Overheard at the water cooler: "Did you hear? There's a music app that makes songs so fast, even my teenage son is starting to feel outdated!"

By the numbers: - 30% of the top 100 generative AI apps popped up in the last six months. - 52% of these apps are all about content generation; think images, music, and even video.

The bottom line: The AI revolution isn’t knocking; it's knocking down doors!

If you want to stay relevant in this tech-savvy world, keeping an eye on emerging tools and learning how to leverage them could spell the difference between business as usual and business on steroids.

So, roll up your sleeves, dive in, and who knows? Your next great idea could be one algorithm away!

[mariacallous]

Lena Anderson isn’t a soccer fan, but she does spend a lot of time ferrying her kids between soccer practices and competitive games.

“I may not pull out a foam finger and painted face, but soccer does have a place in my life,” says the soccer mom—who also happens to be completely made up. Anderson is a fictional personality played by artificial intelligence software like that powering ChatGPT.

Anderson doesn’t let her imaginary status get in the way of her opinions, though, and comes complete with a detailed backstory. In a wide-ranging conversation with a human interlocutor, the bot says that it has a 7-year-old son who is a fan of the New England Revolution and loves going to home games at Gillette Stadium in Massachusetts. Anderson claims to think the sport is a wonderful way for kids to stay active and make new friends.

In another conversation, two more AI characters, Jason Smith and Ashley Thompson, talk to one another about ways that Major League Soccer (MLS) might reach new audiences. Smith suggests a mobile app with an augmented reality feature showing different views of games. Thompson adds that the app could include “gamification” that lets players earn points as they watch.

The three bots are among scores of AI characters that have been developed by Fantasy, a New York company that helps businesses such as LG, Ford, Spotify, and Google dream up and test new product ideas. Fantasy calls its bots synthetic humans and says they can help clients learn about audiences, think through product concepts, and even generate new ideas, like the soccer app.

"The technology is truly incredible," says Cole Sletten, VP of digital experience at the MLS. “We’re already seeing huge value and this is just the beginning.”

Fantasy uses the kind of machine learning technology that powers chatbots like OpenAI’s ChatGPT and Google’s Bard to create its synthetic humans. The company gives each agent dozens of characteristics drawn from ethnographic research on real people, feeding them into commercial large language models like OpenAI’s GPT and Anthropic’s Claude. Its agents can also be set up to have knowledge of existing product lines or businesses, so they can converse about a client’s offerings.

Fantasy then creates focus groups of both synthetic humans and real people. The participants are given a topic or a product idea to discuss, and Fantasy and its client watch the chatter. BP, an oil and gas company, asked a swarm of 50 of Fantasy’s synthetic humans to discuss ideas for smart city projects. “We've gotten a really good trove of ideas,” says Roger Rohatgi, BP’s global head of design. “Whereas a human may get tired of answering questions or not want to answer that many ways, a synthetic human can keep going,” he says.

Peter Smart, chief experience officer at Fantasy, says that synthetic humans have produced novel ideas for clients, and prompted real humans included in their conversations to be more creative. “It is fascinating to see novelty—genuine novelty—come out of both sides of that equation—it’s incredibly interesting,” he says.

Large language models are proving remarkably good at mirroring human behavior. Their algorithms are trained on huge amounts of text slurped from books, articles, websites like Reddit, and other sources—giving them the ability to mimic many kinds of social interaction.

When these bots adopt human personas, things can get weird.

Experts warn that anthropomorphizing AI is both potentially powerful and problematic, but that hasn’t stopped companies from trying it. Character.AI, for instance, lets users build chatbots that assume the personalities of real or imaginary individuals. The company has reportedly sought funding that would value it at around $5 billion.

The way language models seem to reflect human behavior has also caught the eye of some academics. Economist John Horton of MIT, for instance, sees potential in using these simulated humans—which he dubs Homo silicus—to simulate market behavior.

You don’t have to be an MIT professor or a multinational company to get a collection of chatbots talking amongst themselves. For the past few days, WIRED has been running a simulated society of 25 AI agents go about their daily lives in Smallville, a village with amenities including a college, stores, and a park. The characters’ chat with one another and move around a map that looks a lot like the game Stardew Valley. The characters in the WIRED sim include Jennifer Moore, a 68-year-old watercolor painter who putters around the house most days; Mei Lin, a professor who can often be found helping her kids with their homework; and Tom Moreno, a cantankerous shopkeeper.

The characters in this simulated world are powered by OpenAI’s GPT-4 language model, but the software needed to create and maintain them was open sourced by a team at Stanford University. The research shows how language models can be used to produce some fascinating and realistic, if rather simplistic, social behavior. It was fun to see them start talking to customers, taking naps, and in one case decide to start a podcast.

Large language models “have learned a heck of a lot about human behavior” from their copious training data, says Michael Bernstein, an associate professor at Stanford University who led the development of Smallville. He hopes that language-model-powered agents will be able to autonomously test software that taps into social connections before real humans use them. He says there has also been plenty of interest in the project from videogame developers, too.

The Stanford software includes a way for the chatbot-powered characters to remember their personalities, what they have been up to, and to reflect upon what to do next. “We started building a reflection architecture where, at regular intervals, the agents would sort of draw up some of their more important memories, and ask themselves questions about them,” Bernstein says. “You do this a bunch of times and you kind of build up this tree of higher-and-higher-level reflections.”

Anyone hoping to use AI to model real humans, Bernstein says, should remember to question how faithfully language models actually mirror real behavior. Characters generated this way are not as complex or intelligent as real people and may tend to be more stereotypical and less varied than information sampled from real populations. How to make the models reflect reality more faithfully is “still an open research question,” he says.

Smallville is still fascinating and charming to observe. In one instance, described in the researchers’ paper on the project, the experimenters informed one character that it should throw a Valentine’s Day party. The team then watched as the agents autonomously spread invitations, asked each other out on dates to the party, and planned to show up together at the right time.

WIRED was sadly unable to re-create this delightful phenomenon with its own minions, but they managed to keep busy anyway. Be warned, however, running an instance of Smallville eats up API credits for access to OpenAI's GPT-4 at an alarming rate. Bernstein says running the sim for a day or more costs upwards of a thousand dollars. Just like real humans, it seems, synthetic ones don’t work for free.

[galtechlearninghub]

Ever wondered how to create a functional Todo List web app? 🤔 Dive into our latest blog where we guide you step-by-step using the powerful Claude AI! From coding to deployment, we’ve got you covered. 💻✨

🔗 Read Now