April has been a fantastic month for AI news. Between new open source programs, to Microsoft upsetting Google in the search engine space. AI has yet again proven to be an impactful disruptor in multiple industries. So let’s take a dive into some of the top pieces of AI news we saw in April.
It’s no surprise that AI has become a major tool in the medical field. Thanks to its natural ability to sort data and detect patterns often missed by human medical professionals. It’s fast becoming the “go-to” tool for those who wish to help advance the field of medicine. Last month, an 11-year-old programming prodigy named Leena Rafeeq announced the creation of a new AI app that is able to detect eye diseases simply by using a phone.
Though currently limited to iPhones, this app uses machine learning algorithms and advanced computer vision to analyze various parameters to detect specific diseases with a 70% accuracy rate.
With the rapid advancements of AI, it was only a matter of time before governments around the world sought out ways to create legal frameworks to protect the interests of their citizens. But a proposed EU law has the Python Software Foundation concerned due to possible effects on the open-source community.
According to a new paper by researchers out of Sandford, they were able to create an entire Sim-like town using generative AI. Not only that, but each avatar, powered by ChatGPT, was able to produce unique identities, memories, and behaviors. All of which interacted independently of each other. Of course, this isn’t General artificial intelligence, it’s still important research that allows for the creation of more powerful programs that can interact, and learn with minimal human contact.
In a new advancement in the costume large language model market, Dolly 2.0 provides a list of new features that is sure to impress those looking for their own LLM. According to their post, it is capable of following instructions, enabling organizations to build, own and customize LLMs for their specific needs. In short, if a company wants to use an LLM for sentiment analysis of customer reviews, they don’t have to start from the foundations. With Dolly, they could start with a pre-trained LLM and fine-tune it on a data set of customer reviews.
To anyone watching the rivalry between the two juggernauts of the data industry, Google has been attempting to catch up with Microsoft-backed OpenAI for a few months now. With Bing’s AI enhancements exploding in popularity, this has forced Google to push forward with their own AI-powered search tools. But this won’t be a single drop of features, Google, according to reports, will be spending the remainder of 2023 launching AI-powered features to its user base.
SpaceX/Tesla owner, Elon Musk, announced his plans to launch his own rival chatbot to compete against OpenAI’s ChatGPT. According to Musk, TruthGPT’s goal is to be transparent, something that he has openly been critical of OpenAI for some time. To him, AI safety is a major concern that not enough tech leaders are paying attention to.
The use of web scraping and other data for training purposes has put many companies and users on edge. Many feel that by allowing AI companies to use their data, often without consent, to train their models, is tantamount to plagiarizing. So popular developer reference website, Stack Overflow has taken an interesting approach. AI companies will now how to pay to use data from the website. In turn, the company will reinvest the revenue into the community.
As mentioned in a previous story highlight, Google has been struggling to catch up with Microsoft in the generative AI market; especially since ChatGPT’s popularity exploded last year. So to facilitate this, Google announced that they were going to merge their teams in DeepMind and Brain into a single team named Google DeepMind.
Stability AI has introduced StableVicuna, the first large-scale open-source chatbot trained using reinforcement learning through human feedback (RLHF). It is an instruction fine-tuned and RLHF-trained version of Vicuna v0 13b, a chatbot model. StableVicuna has undergone benchmarking against other similar open-source chatbots and has demonstrated strong performance.
The development of StableVicuna was made possible through the contributions of Duy Phung, open-source contributors, and datasets from OpenAssistant, Anthropic, and Stanford, with acknowledgment given to OpenAssistant’s team for providing early access to the RLHF dataset.
According to reports, Microsoft is holding the edge in AI against its main rival, Google. Over the last several months, Microsoft has been quite open about its push to use generative AI as a means to gain greater market share; particularly in markets that have been dominated by rivals such as Google.
Pretty neat right? New news is coming out every day related to AI, as 2022, became the dawn of AI for much of society, 2023, is fast becoming the foundation of AI and its possible effects on humanity in the long term. Stay tuned to our next Top AI News round-up in June!