Innovative AI - Artificial Intelligence (AI) and its impact on businesses.
Google's AI Evolution: From Search Algorithms to Technological Mastery -A Chronicle of Google's AI Endeavors
This week in Innovative AI
Innovative AI Shorts:
Unlocking the Potential of Language Models in Healthcare: Challenges and Opportunities
Collaboration is Key: Tackling Fraud in the Age of AI and Real-Time Payments
New Study Reveals Cybersecurity Threat: Typing Passwords on Zoom May Lead to eavesdropping attacks
Executives’ Highlight: Google's Journey in AI: A Comprehensive Exploration of Technological Advancement
Prompts for your Organization: ChatGPT for creating Midjourney prompts; ChatGPT for creating Midjourney prompts
Overview: Center for Deep Tech Innovation Events
Tools to try out: Heuristica
Innovative AI Shorts
Unlocking the Potential of Language Models in Healthcare: Challenges and Opportunities
Language models like OpenAI's ChatGPT and Google's PaLM have the potential to streamline administrative work and answer medical questions in patient portals, but integrating them into clinical workflows presents challenges. A recent article in Nature shows that although Google's Med-PaLM delivered accurate answers to medical questions, it fell short compared to clinicians' responses. Questions surrounding patient privacy and safety need to be addressed before the widespread use of these models in clinical applications. Despite this, the field is rapidly advancing, with new models like Med-PaLM 2 surpassing previous performance on medical questions. The study highlights the need for benchmark questions and ongoing research to determine the best ways to combine human expertise with AI technology in healthcare decision-making.
Collaboration is Key: Tackling Fraud in the Age of AI and Real-Time Payments
The rise of easy-to-access artificial intelligence (AI) tools and real-time payments in the US has led to an influx of fraudsters taking advantage of the system. This development poses significant implications for both consumers and businesses, raising questions about liability and how to mitigate fraudulent activity. In response, industry consortiums, such as SardineX, Beacon, and efforts led by Mastercard, have emerged to tackle payment-related fraud. However, there is a need for greater collaboration among these consortiums to ensure the sharing of crucial data and the development of effective solutions. The regulatory landscape and technological challenges surrounding data storage and privacy also complicate fraud prevention efforts. To address this, a comprehensive defense strategy incorporating AI and involving both the private and public sectors is crucial. Without collaboration, progress in combating fraud risks being undermined.
New Study Reveals Cybersecurity Threat: Typing Passwords on Zoom May Lead to eavesdropping attacks
New research highlights the potential security risks of typing in computer passwords during online video calls. The study suggests that artificial intelligence can identify which keys are being pressed on a laptop keyboard by eavesdropping on the sound of typing. The researchers used machine learning algorithms to create a system that can accurately determine which keys are being pressed with over 90% accuracy based on sound recordings. With the increased use of video conferencing tools and devices with built-in microphones, the threat of cyber-attacks based on sounds has also risen. Experts emphasize the need for public discussions on the governance of AI, as well as the adoption of mitigation measures such as biometric passwords or two-step verification systems.
Google's Journey in AI: A Comprehensive Exploration of Technological Advancement
Google's evolution from a simple search engine to a global technological powerhouse is paved by innovation and resulting growth. A pivotal aspect of this growth trajectory has been its strategic investment in machine learning/Artificial Intelligence (AI).
Google's maiden venture can be traced back to its foundational product, the search engine itself. The PageRank algorithm, responsible for ranking web pages based on relevance and significance, marked Google's initial foray into intelligent algorithms. With the progression of the digital era, Google's algorithms subsequently underwent continuous refinement, incorporating AI and machine learning techniques to enhance search accuracy and advertisement optimization.
Recognizing the extensive potential of AI, Google launched TensorFlow in 2015. This open-source machine learning platform empowered developers across the globe to explore and implement AI solutions. The democratization of AI development through TensorFlow led to a proliferation of AI applications across diverse sectors. Apart therefrom the acquisition of DeepMind, a leading AI research organization based in London in 2014, has to be regarded as one of the major milestones in Google’s AI journey. This move not only underscored Google's earnest commitment to AI but also set the stage for groundbreaking innovations. One of the best-known examples, thereto, is DeepMind's AlphaGo, the AI model that triumphed over the world champion Lee Sedol in the complex board game Go in 2016 (Go is a two-player board game considered to be the most complex of all strategy games known in the world).
Also in 2016, Google introduced Google Assistant, marking a shift in AI's accessibility, transcending the confines of research labs. Empowered by AI, Google Assistant facilitated voice recognition and natural language processing across various devices, including smartphones and smart speakers. This innovation enabled users to execute tasks, retrieve information, and manage smart devices through simple voice commands.
Since then Google has been integrating AI into its product portfolio wherever it was possible. Google Photos leverages AI for tasks such as image recognition and categorization, enhancing user experience. Google Translate, equipped with real-time translation functionalities, serves as a daily tool to overcome language barriers (although we believe that DeepL still achieves better results when translating texts in different languages). Google Maps employs AI to analyze real-time traffic data, providing users with optimal routing solutions.
Recent developments in Google’s AI journey.
Google has been (further) developing many AI initiatives, most of them being introduced during Google’s annual developer conference Google I/O. As for instance, Google Maps has been revolutionized by AI, offering users eco-friendly routes, safer options, and personalized map details. Google Translate has introduced "Transcribe" in eight languages, facilitating real-time translations for events like lectures or interviews.
Chatbots and Beyond:
One of the main focuses of Google’s developments towards AI still remains in the development of chatbots and beyond.
Google Bard, Google’s AI-driven chatbot, is one of Google's innovative ventures into the realm of AI. Initially grounded in the Large Language Model, LaMDA, its foundation transitioned to PaLM 2 by May 2023. This development was catalyzed by the notable success of OpenAI's ChatGPT. Launched in a restricted capacity in March 2023, Bard's availability gradually expanded to a broader audience.
Drawing inspiration from literary history, the chatbot is christened after William Shakespeare, colloquially known as the "Bard of Avon".
In terms of functionality, Bard mirrors the conversational capabilities of ChatGPT, adeptly addressing a diverse array of queries. It harnesses online databases to furnish users with authentic and insightful responses. Underpinning Bard's operations are the Transformer neural network and Google's proprietary language model, LaMDA. Post its preliminary release, access to Bard was extended to users in the USA and the UK through a waitlist system. A significant announcement at Google I/O 2023 revealed Bard's expansion to over 180 countries, albeit with an exclusion of European Union nations - potentially due to GDPR compliance intricacies. This limitation was subsequently lifted by July 13, 2023, marking Bard's pan-European availability. However, as most of you may know, Bard's journey wasn't without challenges. A notable incident involved the chatbot disseminating an erroneous answer related to the James Webb Space Telescope, resulting in a substantial dip in Alphabet's market valuation.
In addition to the recent developments around Google Bard, Google is also in the phase of developing and testing Gemini - Google’s next-generation foundation model. The name "Gemini" stands for "Generalized Multimodal Intelligence Network," referring to its ability to process and interpret a variety of data formats, including text, images, and 3D models. This project, driven by industry leaders such as Sergey Brin and renowned teams such as Google Brain and DeepMind, aims to push the boundaries of what AI chatbots can do by integrating image generation technologies, therefore, aiming to provide users with a comprehensive and multimodal source of information.
Find more on Googles Gemini along with further developments on Google’s AI developments in Sundar Pichai, CEO of Google and Alphabet, opening speech at Google’s annual developer conference: Google I/O (Information on Gemini 13:30)
Google's Search Generative Experience (SGE)
Another endeavor from Google is its Search Generative Experience (SGE), aiming to refine users’ search experience. This feature, soon to be on Chrome for desktops, offers concise summaries of extensive content. Once activated, users will be presented with AI-generated summaries, enriched with images and the possibility to ask follow-up questions to enhance the search results. Accordingly, SGE is intended to simplify the search for information and provide users with concise content. Enhancing the search experience, therefore, marks a clear strategic step for Google to defend its market segment against other (upcoming) AI tools targeting internet searching.
The Future of Google's AI Initiatives:
Google's vision for AI is expansive, with plans to introduce generative AI tools across select web platforms, potentially offering AI-generated article summaries. With innovations like SGE and Gemini, Google envisions a future where AI is an integral part of our daily lives. Recent strategic moves, such as the merger of DeepMind with Brain, highlight Google's accelerated AI pursuits. The combined entity is exploring the potential of generative AI as a personal life coach, with Google DeepMind delving into a myriad of tasks, from life advice to tutoring.
All of these endeavors underscore Google's urgency to lead in AI and its growing trust in AI systems for nuanced tasks. Nevertheless, Google is also a strong advocate for the responsible use and development of AI. Therefore, Google has established principles for developing AI that have been guiding the company since 2018 - emphasizing responsibility and purpose-driven innovation.
These Principles are the following:
Be socially beneficial.
2Avoid creating or reinforcing unfair bias.
Be built and tested for safety.
Be accountable to people.
Incorporate privacy design principles.
Uphold high standards of scientific excellence.
Be made available for uses that accord with these principles.
In conclusion, there are many interesting (further) AI developments to come from Google. Some of these will target Google's traditional business segments, while others will be in segments you wouldn't necessarily expect Google to be in, such as mitigating the environmental impact of the airline industry by preventing contrails (see our last issue).
We will certainly keep an eye on all of these developments and keep you informed.
Prompts for your Organization
ChatGPT for creating interview questions
“Create five questions for an informational interview with a successful entrepreneur to learn more about their journey and tips for success.”
ChatGPT for creating Midjourney prompts
“Pretend you are creating an allegorical Painting painting on the use of AI to enhance or destroy human civilization. Generate five prompts that AI art tools like Midjourney can use to generate visuals for the painting.”
Overview: Center for Deep Tech Innovation Events
As some of our readers might know, we offer a series of webinars on the topic of AI (and other technologies in the near future).
Below you find a list of upcoming webinars:
Wednesday, September 6, 2023, 4:00 p.m. (CET) - Prompt Engineering for Business Transformation: A Workshop
Wednesday, October 4, 2023, 4:00 p.m. (CET) - Prompt Engineering for Business Transformation: A Workshop
more to come!
For more content, feel free to visit our YouTube channel and LinkedIn
Try-it-out: Heuristica
Heuristica is an AI-powered platform for content generation on a wide range of topics. Primarily, it is a tool that allows users to research a chosen topic or access pre-researched content. Users can thereby use connectors to learn more about a topic without having to write prompts. Heruistica is accessible via web-enabled devices and is quite intuitive to use. However, the outcomes of the searches are usable for private use at best, as the prompts applied seem to be quite basic. Still, the platform is a good resource for people who want to easily search for information on various topics and have it displayed graphically ( though the possibility of hallucinations, etc. is always present).
Here is a short excerpt from our try-it-out:
ChatGPT for creating Midjourney prompts
ChatGPT for creating interview questions
