Hey Creators!

Welcome back to this exciting week with Generative AI, where X users can now share AI-generated NSFW content with the platform's permission, and LLM-powered search engines are about to disrupt traditional search engines. DuckDuckGo just scored brownie points with those who value their privacy with their latest AI offerings. Adobe issued a clarification following backlash from users due to its latest terms and conditions. Stable Diffusion 3 is almost here, but those who got early access are already bringing interesting ideas to life. With LLMs learning to be deceitful and lying, where does that leave us with the trust factor when it comes to AI? In order to test out which LLMs are actually giving correct responses as opposed to which are being deceitful, why not try out all premium LLMs like GPT-4, Claude Opus, Mistral Large, and GeminiPro all in a single conversation window? Try out all premium chatbots in a single window to know for yourself. Try For Free

What’s Cooking Today

• LLMs Have Started Being Deceitful and Have Learned to Lie

• Other Newsletters We Recommend

• Adobe Issues Clarification Amid Backlash From Pro Users

• DuckDuckGo Launches Platform to Access AI with Privacy

• X (Formerly Twitter) Now Allows Users to Share NSFW Content

• Search Engines Are Set to Change with AI and LLMs, But How?

• Fun with SDXL 3

Other Newsletters we recommend

Geekout Newsletter

Let's keep it short and invite you to The World's Most Useful Newsletter for Social Media Managers (30k+ subscribers). Check it out here. 

Creator Logic

The second recommendation is a newsletter that has helped us much in monetizing through digital assets and social media.  Check out their post on the creator monetization framework in the featured section here.

How Much To Trust Deceitful LLMs ?

Be careful when eliciting responses from LLMs; recent studies have found LLMs to be quite deceptive and to have exhibited signs of lying. German AI ethicist Thilo Hagendorff goes so far as to say that sophisticated LLMs can be encouraged to elicit "Machiavellianism," or intentional and amoral manipulativeness, which "can trigger misaligned deceptive behavior." GPT-4, for instance, exhibits deceptive behavior in simple test scenarios 99.16% of the time, according to the University of Stuttgart researcher citing his own experiments in quantifying various "maladaptive" traits in 10 different LLMs, most of which are different versions within OpenAI's GPT family. The model engaged in premeditated deception, broke the deals to which it had agreed, and told outright falsehoods. Keep Reading

Does Adobe Own Your Creations?

Adobe backtracks after a frenzy of backlash over their terms and conditions. The company initially issued a rather dismissive statement that its terms have been in place for years and it was merely clarifying them, but subsequently wrote a blog post which tackled the issue in more detail. Apparently, pro users claimed they were forced to agree to new terms, which included some frightening-sounding language. It seemed to suggest Adobe was claiming rights over their work. When we requested comment from Adobe, the company's initial statement didn't really help, thanks to a dismissive 'nothing to see here, move along' tone. However, the company did subsequently realize the issue wasn't going to go away until it offered a proper explanation. Know More

DuckDuckGo’s Privacy Boosted AI Platform

DuckDuckGo, the search engine for those who value their privacy, has just launched its own AI service which prides itself on the privacy it provides. The company's commitment to privacy is a direct response to the growing concerns about data collection and usage in the AI industry. By providing a private and anonymous platform for users to interact with AI chatbots, DuckDuckGo is setting a new standard for the industry. Users can choose from four AI models: two closed-source models and two open-source models. The closed-source models are OpenAI's GPT-3.5 Turbo and Anthropic's Claude 3 Haiku, while the open-source models are Meta's Llama-3 70B and Mistral AI's Mistral 8x7b. Know More

X Now Allows Users To Share AI Generated NSFW Content

X (formerly Twitter) has updated its rules to formally give permission to users to share consensually-produced AI-generated NSFW content, provided it is clearly labeled. This change aligns with previous experiments under Elon Musk's leadership, which involved hosting adult content within specific communities. While X has never outright said "no" to porn by banning it, these new clauses could pave the way for developing services centered around adult content, potentially creating a competitor to services like OnlyFans and enhancing its revenue streams. This would further Musk's vision of X becoming an "everything app," similar to China's WeChat. Know More

New Era For Search Engines In 2024

If you thought that search engines were going to remain the same in the era of LLMs, then you stand corrected. Imagine a world where search engines understand your queries as a human would, providing answers that are not just relevant but insightful and deeply contextual. This is the world Large Language Models (LLMs) like GPT-4 and their open-source counterparts are ushering in.

Integrating Large Language Models (LLMs) into search engines enhances their ability to understand and respond to user queries by providing more accurate and contextual responses. To further improve search engines, continuous model training on new data, incorporating user feedback, and personalizing search results based on user behavior are essential steps. Additionally, integrating multilingual voice search capabilities makes the search experience more accessible and natural for users worldwide. This involves converting spoken queries into text using Automatic Speech Recognition (ASR) systems, processing the text with the LLM, and converting the LLM's response back to speech using Text-to-Speech (TTS) systems. Know More

Fun With SDXL 3