The Good News of Enhanced Search with Cynthia

UKTN, the UK's premier online media brand for the technology industry, has firmly established itself as an indispensable resource for comprehensive coverage of the nation's vibrant tech ecosystem. From fledgling startups to rapidly growing scaleups and established tech unicorns, UKTN's journalism spans the length and breadth of the United Kingdom. UKTN places innovation and entrepreneurship at the heart of its coverage, meticulously tracking funding news, analysing industry trends, profiling groundbreaking startups, and sparking engaging conversations with expert opinions. At Cynthia Systems, we are ardent followers of UKTN's content, relying on their astute insights to stay ahead of the curve in the fast-paced world of technology, with a keen eye on the transformative potential of artificial intelligence and the vital importance of renewable energy solutions.

Reviewing AI and Renewable Energy in 2024 and Beyond

At Cynthia Systems, we find ourselves captivated by the UKTN website and the exceptional readership experience it offers. The refreshing green hues that permeate the site resonate deeply with our sensibilities, evoking notions of freshness, lightness, and an unwavering commitment to ecological awareness. Just as the vibrant green of springtime breathes new life into the world after the dormancy of winter, these colours symbolise the sprouting of our hopes and dreams, made manifest through our tireless efforts.

In the spirit of fostering a dynamic growth mindset and revitalising our ideas for 2024 and beyond, we embarked on a journey of discovery using UKTN's search functionality, focusing our attention on the realms of artificial intelligence and renewable energy. Our aim was to unearth the hidden gems within the United Kingdom's thriving AI community—those forward-thinking companies that embody a modern and progressive outlook on sustainability, responsible stewardship, and green practices.

Elevating UKTN's Search Engine with Cynthia

At Cynthia Systems, we constantly challenge ourselves to push the boundaries of what's possible, asking, "How does Cynthia take it to the next level?" This core mantra is deeply ingrained in our approach to designing solutions and building new models, as we firmly believe that true leaders consistently raise the bar on the quality, beauty, and excellence of their products and services.

With this in mind, we set Cynthia to work, tasking her with crawling UKTN's extensive library of 11,867 articles on Friday, April 30th, 2024. We then took the process a step further by enhancing UKTN's content with Generative AI, leveraging Cynthia's partnerships in the AI ecosystem and our proprietary library of prompt engineering formulas, meticulously optimised for British tone, grammar, and spelling. To ensure optimal performance, we trained Cynthia's UKTN search model on a cluster of 8 x A100, 80 GB SXM4 NVIDIA GPUs, equipped with 240 vCPUs, 1.8 TiB RAM, 20 TiB SSD, and powered by PyTorch Version 2.0.1.

Measuring Page One Precision in UKTN's Search Results

When querying UKTN's built-in search engine for the latest AI news using the phrase 'AI 2024', one is presented with a substantial volume of results spanning 35 pages. While the first result, titled "Industry predictions for AI in 2024", appears highly relevant, further examination of the first page reveals several articles that, at first glance, seem to bear little connection to the world of AI.

Titles such as " From Tony Blair to Monzo, key players demand UK fintech reform", " BT aims to double digital skills team by 2024", " Skyports secures £88m to build Dubai flying taxi service", and " Scottish rocket firm Orbex raises £16.7m to fuel upcoming launch" populate the search results, leaving readers to question their pertinence to the initial query. Upon closer inspection, it becomes evident that only 8 out of the 12 results on the first page are indeed relevant to AI in 2024, while the remaining 4 prove difficult to interpret in terms of their applicability.

This translates to a precision rate of 66%, a common metric in machine learning that assesses the proportion of relevant results returned by a search engine. Consequently, 33% of the results were deemed imprecise by human judges, highlighting the need for a more refined and targeted approach to information retrieval.

Cynthia's Laser Sharp Precision for UKTN's Search Experience

In a remarkable demonstration of AI-powered precision, Cynthia's search engine delivers laser-sharp results for the query 'AI 2024' on the UKTN platform. The superiority of Cynthia's performance is immediately evident, with the generated search result previews standing in stark contrast to the sparse and often incomplete snippets provided by UKTN's current grid layout and Google's staccato keyword-matched fragments.

While these conventional approaches may prioritise search result density at the expense of precision, Cynthia's previews offer clear, insightful, and concisely summarised information, enhanced by explainability highlights. This exceptional performance is a testament to Cynthia's dual machine learning objective function, which elegantly balances relevance and ease of confirmation, respecting the reading order of both left-to-right languages like English and French, and right-to-left languages such as Arabic and Farsi.

The efficacy of this approach is underscored by Cynthia's impressive 96% precision across 50 search results, with only two partially relevant articles appearing at positions 31 and 42. This showcases Cynthia's unparalleled ability to deliver semantically and positionally intelligent search results, setting a new standard in the realm of AI-driven information retrieval.

At results 48, 49, and 50, Cynthia continues to return relevant and meaningful results, enhanced by insightful explainability and engaging Generative AI previews. The Cynthia REST API reveals that Cynthia identifies a total of 94 results she deems relevant for the query 'AI 2024'. Cynthia offers two primary options to control the number of results returned to the searcher: an upper limit on the number of results, set to 50 in the Cynthia Playground in this example, and a feature called 'auto-limit'. With 'auto-limit', Cynthia selectively truncates the results to include only those she considers truly relevant and worthy of your time and attention.

The Long-tail of Search and the Business Case for Cynthia

Search engine optimisation (SEO) professionals classify search queries into three main categories based on their popularity and specificity: the "head," the "body," and the "long-tail." Head terms, also known as "short-tail," are often one or two-word search phrases with ambiguous intent, while body phrases are slightly more specific and usually contain more words, providing a clearer idea of the searcher's intent.

Long-tail queries, which generally contain three or more words, illustrate a more specific intention from the searcher. These long-tail queries comprise the largest number of unique searches and boast the highest conversion rates, as users who issue them know what they want and are more likely to engage with long-tail content and make purchases with greater certainty.

Cynthia revolutionises the world of SEO by automatically capturing the long-tail of search queries in a completely automated manner, eliminating the need for expensive data labelling or annotation in low-cost labour countries. Cynthia's design, based on self-supervised learning of the web, similar in size and scope to GPT-4's training dataset but focused on insightful comprehension rather than chatbot productions, enables her to target long-tail keywords accurately and build a website's authority. This makes Cynthia orders of magnitude better, cheaper, and more ethical than all existing AI search and discovery systems on the market today.

Measuring UKTN's Ability to Retrieve Long-tail Results

To demonstrate Cynthia's ability to service the long-tail of search queries, we at Cynthia Systems, as thought leaders in renewable energy and with our founder's Scottish descent, crafted the query 'Renewable energy Scotland 2024' and reviewed UKTN's current ability to service the long-tail of search. The results from UKTN were scarcely populated, with only two results: " Scottish rocket firm Orbex appoints CEO and chair" and " Berkshire to home BP-backed EV battery research facility".

The first result is clearly about Scotland, but unless one knows the details of the R&D ambitions of Orbex, including their future aspirational development of biofuels for rocketry, it is unclear why they are relevant to renewable energy, especially since rockets are highly polluting, let alone the fact that the vacuum of space is a lamentable wasteland where energy and resources would need to be sent in unsustainable quantities to support human space travel.

The second result is clearly relevant to renewable energy, regarding the future of battery technology for electric vehicles; however, Berkshire is a ceremonial county in South East England, totally separated from Scotland. Please note that neither of these two results is about the current state of renewable energy in 2024. If one did not know better, one might be inclined to surmise that UKTN simply has no content relevant to renewable energy in Scotland in 2024; however, this is certainly not the case, and Cynthia will now prove it.

Cynthia's Mastery of Long-tail Queries via Deep Learning

Cynthia's mastery of the long-tails starts in position 2, where we see that Piclo, the UK's first peer-to-peer renewable energy trading platform, launched in collaboration with cleantech innovator Open Utility and renewable supplier Good Energy. The platform has already signed up a number of customers, including Cornwall's Eden Project botanical gardens and several National Trust properties across England, Scotland, and Wales.

Next, in position 7, we see the pioneering EV battery maker to build a 'megafactory' in Scotland. AMTE Power, a London-listed battery maker, is set to construct an EV battery "megafactory" in Dundee, Scotland, which will develop lithium-ion and sodium-ion batteries essential for manufacturing electric vehicles.

In positions 12-14, we see that Cynthia understands both meronyms and holonyms, specifically that Glasgow is a constituent part of Scotland, in addition to synonyms and related terms, such as Scottish and electricity. With the nuanced understanding that hydrogen, wave power, low-carbon and sustainable practices are all a vital part of renewable energy, Cynthia recognises and features both Clyde Hydrogen and Mocean Energy.

Next in positions 23-24, we see additional confirmation that Cynthia understands city, county, state and regional information, highlighting collaborations between Edinburgh and Wales through Nova Innovation, in addition to low-carbon AI supercomputing at the University of Edinburgh.

In position 42, we see that Edinburgh-based startup Gravitricity has secured nearly £829,000 from 1,085 investors in a crowdfunding raise on Crowdcube, exceeding its initial goal of £550,000. The funds will be used to hire staff and develop infrastructure as the company progresses its gravity-harnessing method of generating electricity and below-surface green hydrogen storage.

Finally in position 44, we see the resurgence of wave power with an earlier story about Edinburgh's Mocean Energy funding from the EU. Mocean Energy, an Edinburgh-based wave energy machine company, has secured £3.2m in EU funding to deploy their Blue Horizon 250 (kW) device in Orkney.

All of the results were relevant to renewable energy in the UK, and many of them demonstrated pinpoint accuracy in recognising the thought leadership in environmental sustainability and renewables coming from Scotland. These results also showcase Cynthia's explainability approach to highlighting relevant terms and phrases in her Generative AI previews, which Mekka Okereke of Google explains is a billion dollar feature, now adapted to Cynthia within her deep neural network activations.

Cynthia is Sustainable AI for the United Kingdom and Beyond

Cynthia Systems, a pioneering startup conceived in Edinburgh, Scotland during the level-5 lockdown of January 2021, is revolutionising online experiences with Cynthia, our groundbreaking AI search engine who transforms e-commerce, knowledge discovery, and hyper-targeted contextual advertising. Born in the Mayvn Cafe at Eden Locke, Cynthia solves the problem of irrelevant search results through Lockean empiricism.

“Reading furnishes the mind only with materials of knowledge; it is thinking that makes what we read ours.” ―John Locke, The Common School Journal (1843)

Our growth trajectory includes exclusive UK partnerships, case studies, visibility, international expansion and growth, followed by a public IPO on the London Stock Exchange, aspiring to join the FTSE 100. Connect with us today to discuss our employment opportunities along with our exclusive HR partner, Tarento Global, our influencer programmes, and integrating with our beloved Synthetic Intelligent Agent™, Cynthia.

Follow us on LinkedIn to stay up to date on Cynthia's stunning entrance into the United Kingdom, where she is set to revolutionise the future of search, discovery and contextual advertising in a manner that is socially responsible and environmentally friendly for everyone.