Development of multimodal interactive systems is hindered by the lack of rich, multimodal (text, images) conversational data, which is needed in large quantities for LLMs. Previous approaches augment textual dialogues with retrieved images, posing privacy, diversity, and quality constraints. In this work, we introduce \textbf{M}ultimodal \textbf{A}ugmented \textbf{G}enerative \textbf{I}mages \textbf{D}ialogues (MAGID), a framework to augment text-only dialogues with diverse and high-quality images. Subsequently, a diffusion model is applied to craft corresponding images, ensuring alignment with the identified text. Finally, MAGID incorporates an innovative feedback loop between an image description generation module (textual LLM) and image quality modules (addressing aesthetics, image-text matching, and safety), that work in tandem to generate high-quality and multi-modal dialogues. We compare MAGID to other SOTA baselines on three dialogue datasets, using automated and human evaluation. Our results show that MAGID is comparable to or better than baselines, with significant improvements in human evaluation, especially against retrieval baselines where the image database is small.

Anyone who has recently driven on any portion of Highway 101 between San Jose and San Francisco knows that our roads are increasingly crowded, largely because of the enormous number of people working for Silicon Valley tech companies. Apple, Google, Facebook and Cisco account for about 80,000 local workers, according to Silicon Valley Business Journal, and that doesn't count the thousands more people who work at Microsoft, Tesla, Linked-In, Oracle and other large companies plus the many local startups. And, with a significant number of tech workers choosing to live in San Francisco and the South Bay, traffic flows in both directions. Plus, the Bay Area's staggering home prices have forced many workers, especially those who don't earn six-figure salaries, to live in outlying communities as far away as Modesto where 7.3 percent of workers travel at least 3 hours a day to get to and from work – many to Silicon Valley. Stockton is even higher at 10 percent, according to a 2019 Apartment List study.

Netflix moochers beware, you may have to start paying for your own account. At a technology event in Las Vegas, a software firm Synamedia unveiled their Orwellian AI system which has been created to track down any account that has been sharing login information with another. The software works by analyzing geolocation data to determine which accounts are logged in at any one time and from where. By doing so they can see who is sharing their credentials and where. It can even decipher your location type like your home, work or if you are accessing from a hotel or vacation location.

Freely sharing Netflix login details with friends and family who live minutes, hours, or days away could become a thing of the past. The company is introducing new artificial intelligence (AI) technology to identify users who share login details with others to avoid paying an additional fee for online entertainment services. "The Netflix service and any content viewed through our service are for your personal and non-commercial use only and may not be shared with individuals beyond your household," Netflix's terms and conditions state on its website. Magid conducted video entertainment survey that found 26 percent of millennials between the ages of 22 and 40, out of a group of 2,000 Americans between the ages of 8 to 64, admit to sharing account passwords for over-the-top (OTT) services such as Netflix, Hulu, and Amazon Prime Video. Day One at #CES2019 was filled with nonstop customer conversations.

In a previous column I mentioned the Amazon Echo ($179 on Amazon.com) as a useful device for listening to music and podcasts. But, after using the Echo for the past month and hearing others talk about it at CES, I've come to the conclusion that it's far more than that. I bought the Echo as a music player because I was impressed at how easy it is to use your voice to play songs from your own music library that you've uploaded to Amazon Music along with the million or so songs on Amazon Prime Music, your Pandora playlists and the podcasts, music and online radio stations on TuneIn and iHeartRadio. You address the Echo as "Alexa," Amazon's persona that's equivalent to Siri or the OK Google voice commands for Android devices. Amazon also gives you the option to address Echo as "Amazon," but I prefer calling her Alexa.