Join Nick Mishko, Senior Data Analytics Team Lead at Calligo, as he delves into the world of property management analytics and Yardi data.

Discover how Calligo’s data analytics practice transforms Yardi data into powerful tools, enhancing operational efficiency for property management firms globally. From data extraction challenges to creating dynamic dashboards, explore the strategies and solutions that propel businesses forward.

If you’re navigating Yardi complexities or seeking to leverage analytics for your property management endeavours, this insightful discussion is a must-watch. Stay tuned for more insights from Calligo Shorts!

The post Unlocking Property Management Insights: Extracting and Analyzing Yardi Data appeared first on Calligo.

In this lively debate you will hear from Calligo’s Practice Leads as they discuss their key takeaways from 2023 and their data predictions for 2024 and beyond.

Topics discussed include:

The post Data Transformation Predictions for 2024 – Calligo Data Leaders Roundtable appeared first on Calligo.

Artificial Intelligence (AI) and machine learning have revolutionized numerous industries, offering automation and efficiency. However, achieving the optimal balance between human input and machine automation in AI model development is crucial but often overlooked. In our recent Beyond Data podcast, hosts Tessa Jones and Peter Matson were joined for a compelling discussion with the co-founder of Trubrics, Joel Hodgson, where the importance of AI explainability, trust, user feedback, and ongoing monitoring were explored.

The Challenge of Model Adoption

Joel highlighted the challenge of model adoption, a common issue in the data science landscape. Organizations invest significant time and resources in developing AI models, only to face skepticism and underutilization from non-technical stakeholders. This hesitation often arises from a lack of trust and understanding. Education and transparency are vital tools to address this challenge.

Effective Communication and Collaboration

Another significant hurdle is the gap in effective communication between business professionals and data scientists. Bridging this divide is essential to incorporate valuable domain knowledge into the model development process. The solution lies in creating feedback loops that enable collaboration between domain experts, business users, and data scientists throughout the model’s lifecycle. These feedback loops are crucial for gathering user insights, improving model performance, and building trust.

User-Centric Monitoring and Model Utility

Trubrics’ approach of “machine learning monitoring from the users’ point of view” shifts the focus from traditional machine learning metrics to user perception. Evaluating AI models based on their impact and utility to users, rather than just accuracy, is essential. Users’ experiences, trust, and satisfaction play a pivotal role in determining the effectiveness of AI models. Monitoring should identify issues impacting the user experience and ensure AI models align with user expectations.

Building Trust as the Foundation

Trust emerged as a cornerstone in AI adoption. Trust is not limited to data scientists but extends to end-users, employees, and the entire organization. It involves transparent communication, feedback loops, and alignment between different groups. Over time, as individuals become more familiar with AI in the business world, this trust can be built and weaved into organizations’ culture, just as our trust in everyday technology has.

Balancing Technical and Business Monitoring

Monitoring AI models’ performance is essential. Technical monitoring involves tracking various model characteristics, while business-facing monitoring assesses alignment with expectations and business impact. These two facets of monitoring are crucial in ensuring AI models continue to meet user needs and business objectives and therefore must be aligned when identifying the reasoning and desired outcomes from such models.

Measuring ROI and Sustained Value

Measuring and evaluating the Return on Investment (ROI) for AI models presents considerable challenges, especially when examining their performance over extended periods. Striking a balance between the continual expenses associated with model maintenance and the value it delivers requires a nuanced approach. Organizations need to account for both the initial and ongoing financial ROI assessment, recognizing that it can become less clear-cut.

According to a recent research report conducted by Calligo, in collaboration with the Global CIO Institute, “36% of business leaders measure the success of an ML project in financial terms, while 11% either have no way to gauge success or go by gut feeling“. This suggests that determining the ROI for ML and AI initiatives isn’t solely tied to financial gains; it also involves a significant degree of uncertainty when the desired ROI isn’t well-defined at the project’s outset.

In conclusion, AI explainability and the balance between human input and machine automation are crucial in AI model development. Education, transparency, effective communication, user-centric monitoring, and trust-building are essential elements in this endeavor. As AI continues to shape our world, achieving these elements will be pivotal to ensure responsible and ethical AI development and its successful integration into our lives. Organizations like Trubrics are at the forefront of this mission, working towards making AI a valuable and trusted tool in our increasingly automated world.

Listen on Spotify or watch below

The post AI Explainability – Balancing Human-Machine Collaboration and Potential appeared first on Calligo.

In this episode of the Beyond Data podcast series, Tessa Jones (Calligo’s Chief Data Scientist) and Peter Matson (ML Solution Architect) are joined by Martin Hoskin, Chief Technologist at VMware and Advisory Board Member for the Centre for Data Ethics & Innovation. In this enlightening discussion, we delve into the concept of data sovereignty and its implications for ethical data use, as well as explore how federated learning offers a promising solution to the challenges we face. 

Understanding Data Sovereignty

Data sovereignty encompasses the notion of data residency, access control, and governance. The dominance of American cloud providers, subject to U.S. laws, raises concerns about data privacy and security, particularly in the European context. For certain organizations, like government agencies and defense suppliers, data sovereignty becomes a critical factor. VMware has introduced a program to certify partners as Sovereign, ensuring data storage, processing, and governance are specified, differentiating them from major hyperscale cloud providers. 

The Challenge of Data Sharing

Data sovereignty also touches upon the ethical dilemma of sharing data for legitimate purposes like law enforcement investigations. Striking a balance between data privacy and the greater good is complex. For instance, the case of Apple’s cloud security raises questions about when governments should access personal data to combat serious crimes. 

Federated learning emerges as a promising solution to data sharing challenges. This approach enables entities to collaboratively train machine learning models without sharing raw data. Instead, local models are trained on separate datasets, and only aggregated model updates are shared with a central server. This preserves privacy and protects sensitive data, making it suitable for applications like fraud detection in the banking industry. 

Experimenting with Federated Learning

The Centre for Data Ethics & Innovation (CDI) conducted an experiment using federated learning for government-provided services. The CDI set up two data sets—one for detecting fraud in financial transactions using SWIFT data and another for studying the spread of COVID-19. The experiment highlighted the complexities of sharing data, including obtaining government buy-in and ensuring data anonymization to protect privacy. 

While federated learning is ingenious, it comes with its own set of challenges. Concerns arise about the aggregator potentially being reverse engineered to extract sensitive information. Additionally, the scale of data involved in real-world applications may make reverse engineering even more difficult. 

As data continues to play a critical role in various industries, addressing data sovereignty and privacy concerns remains paramount. Federated learning offers a way to enable collaboration without compromising data privacy. However, continuous innovation is necessary to tackle challenges like reverse engineering and fully realize the potential benefits of this approach. 

Ethical Considerations in AI and Data Technology

The conversation takes a broader turn, exploring the intersection of AI, data, and ethics. AI development should consider risks, probabilities, and potential biases to build robust and ethical systems. Ethical implications of sharing genetic data and the responsibility of pharmaceutical companies in handling such information are discussed. 

Regulating AI Ethics and the Divide between Academia and Industry

The need for clear regulations to define and enforce ethical standards in AI and data technology is acknowledged. Balancing philosophical academic perspectives with industry practicality becomes essential as AI progresses toward stronger AI with self-learning capabilities. 

Navigating Legal Frameworks and Data Sharing in Healthcare

Enforcing ethical standards and regulations on a global scale, especially with rogue states, poses challenges. Collaboration through global forums, like Gaia X, can facilitate trust, data security, and individual interpretations of frameworks. Standardized data-sharing frameworks and data portability regulations can address data sharing challenges in healthcare. 

Autonomous Weapons and the Role of Global Forums

The ethical challenges of deploying AI in autonomous weapons, especially in making life and death decisions, raise profound moral dilemmas. The hosts stress the importance of engaging in public discourse and involving the global community to shape AI and robotics’ future. 

The Impact of Social Media on Data Privacy

The podcast concludes with a discussion on the influence of social media on data privacy and the ethical considerations surrounding its use. Addressing the impact on young minds and the potential implications on decision-making, including voting rights for 16- and 17-year-olds, is highlighted. 

In conclusion, data sovereignty, AI ethics, and federated learning are crucial components of an evolving data landscape. Ethical considerations must be at the forefront of AI development and data sharing to ensure responsible and equitable data-driven futures. By embracing ethical practices and fostering interdisciplinary collaboration, we can harness the potential of AI while respecting individual rights and privacy. Establishing global forums and transparent public discussions will play a pivotal role in shaping the future of AI and robotics in a manner that benefits humanity as a whole. 

Listen on Spotify or watch below

The post Data Sovereignty Unveiled – Balancing Rights, Privacy, and Innovation appeared first on Calligo.

Exorcizing the ghost in the machine

In this latest podcast in our ‘Beyond Data’ series, Tessa Jones (Calligo’s Chief Data Scientist) and Peter Matson (Data Science Practice Lead) talk with Oxford University’s Professor Philip Howard about the threats posed to democracy by technology, specifically in the shape of Lie Machines.

Fact or fiction? Microtargeting with lie machines

In this age of social media, chatbots and AI it’s never been easier for individuals to share their opinions.  Instant communication to, and engagement with, a global audience is now commonplace, and it seems there’s no need to let facts get in the way of a good angle. As Mark Twain, or maybe Winston Churchill, or most probably Jonathan Swift famously said, “a lie can travel halfway around the world whilst the truth is still putting on its shoes.” A great example in itself of the ease in which misunderstandings and misappropriations can become canon.

In this vein, Professor Howard has spent years studying the mechanisms in which opinion, behavior and values can be manipulated and misdirected by lie machines:

“Lie machines are large, complex mechanisms made up of people, organizations, and social media algorithms that generate theories to fit a few facts, while leaving you with a crazy

conclusion easily undermined by accurate information. By manipulating data and algorithms in the service of a political agenda, the best lie machines generate false explanations that

seem to fit the facts.”

Lie Machines: How to Save Democracy from Troll Armies, Deceitful Robots, Junk News Operations, and Political Operatives

We find lie machines in all types of countries and governing structures. They share common elements – political actors produce the lies, social media firms distribute them, and paid consultants market them. High profile examples of the effectiveness of the lie machine include the UK’s Brexit campaign, and Trump’s electioneering – in both cases patently untrue ‘facts’ and arguments were targeted at key voters by disinformation networks, troll farms and lie machines. Algorithms direct individuals towards ever-more insular sources and extreme content:

 “A healthy, public-facing algorithm might occasionally introduce another credible source…  we know the platforms play around with this stuff, especially during elections in the US”

Controlled by bad actors and forming a global ecosystem of lie development and propagation, these lie machines spread their tendrils across every social media platform, moving out from Facebook as new outlets develop.

Computational propaganda

Lie machines have evolved and finessed themselves as technology advances. Instead of stealing the photos, social media handles and biographies of real people, AI now generates new pictures and personas and thus evades technology platforms’ troll-spotting software.

Spreading propaganda far and wide, with a convincing voice, the lie machine

• Has a profound effect on society, with a scale that is difficult to quantify
• Is perfectly engineered to target human vulnerabilities, reducing critical thinking
• Deliberately misrepresents and appeals to emotions and prejudices, using our cognitive biases to bypass rational thought and create echo chambers
• Is vague and unknowable – what training data was used for large language models? (Professor Howard postulates that every Gmail sent over the last 25 years may have been scraped, along with content from junk news sites)

Doing better – where does the onus sit? User or developer?

When it comes to developing processes to combat the lie machine, there’s no one legislation or guiding principle that works. We must always consider the regional and cultural context of both data and users. Research can’t necessarily be amalgamated or directly compared from different regions and countries – for example, we know that the placebo effect is always greater in US medical studies. To date, technology has not always built in cultural nuances in how people use words, with intent and meaning lost in translation – the majority of network takedown orders are for sites that are not in English.

Wherever there is human input, there are behavioral differences that make it much more difficult to apply common rules:

“People who manage cookies are above average in terms of their knowledge of technology, so these people are generally more purposeful in terms of how they set up their news feeds and where they go for information”

The huge amount of disinformation spread around Covid and the resulting vaccination campaign demonstrates how potent the lie machine is. It doesn’t need to convince people its argument is right, all that is required is to introduce enough doubt, to highlight there is a chance of harm. After all:

“If everybody really understood probability, nobody would ever buy a lottery ticket”

Balance the field – breaking the lie machines

Professor Howard believes that whilst we are justified in our concern about the threats to democracy, the principles behind the lie machine can be harnessed for good – promoting topics that are in the public interest and generating democratic discourse:

“I am cynical, but not fatalistic”

He describes the steps we can take to break the lie machines:

• Public policy oversight, founded in ongoing public data capture and analysis
• Designing social media to highlight emerging consensus, rather than heated conflict – machine learning can amplify common ground
• Setting election guidelines to create more opportunities for civic expression
• Giving journalists, civic groups and researchers access to all the public opinion data that is currently in the hands of the technology firms
• Ensuring that the big data collected by technology platforms is added to public archives

The answer is more social media, not less. But it needs to serve society much better.

IPIE – bringing down the lie machine

Professor Howard has recently launched a new program, creating an independent scientific body to foster global cooperation in safeguarding the online information environment. The International Panel for the Information Environment (IPIE) will assess the scope of the misinformation crisis, analyze its effects on our societies and the planet itself, and propose solutions. Featuring data scientists and engineers alongside neuroscientists and sociologists, IPIE hopes to be the beginning of a global effort to save our common information environment.

Watch the podcast for yourself below to hear more from Professor Philip Howard about the power of the lie machine, and crucially, to learn how we can use it for the collective good.

Professor Philip Howard is a social scientist with expertise in technology, public policy and international affairs. He is Director of Oxford University’s Programme on Democracy and Technology, a Statutory Professor at Balliol College, and he is affiliated with the Departments of Politics and Sociology. Currently, he is also a Visiting Fellow at the Carr Center for Human Rights at Harvard University’s Kennedy School.

The post Lie Machines – The global fight against misinformation appeared first on Calligo.

In Calligo’s latest Beyond Data podcast, Tessa Jones (Chief Data Scientist) is joined by Dr Ellie Graeden, Research Professor (Center for Global Health Science and Security) at Georgetown University. Here we explore some of the episode’s highlights:

• The inherent conflict of private data and the public good
• Protecting individual rights within federated learning
• The importance of effective communication and a common language
• Designing systems and policies that work together
• Focusing regulation on outcomes, not creating data siloes

At societal level, poor communication costs lives

Transitioning data across and between departments and data systems has historically been fraught with problems – who owns it? Who pays for it? Is it understandable and translatable into meaningful and actionable insights for the end user? 

Having worked extensively in disaster response, Dr Graeden has seen first-hand the potentially life-threatening issues that can arise when government departments’ data platforms produce incompatible outputs:

• If 20,000 people need water, how many pallets need to be shipped?
• If 10,000 electricity meters have been knocked out by a hurricane, how many people need feeding?

In such scenarios, identifying individuals amongst population-level data is crucial if the help provided is to be sufficient.

“We have to be able to really effectively move and communicate and share data that are relevant, in ways that they can get used by people all across the system”

Of course, any data system design should ensure privacy and protection for personal data. ‘Big data’ is still relatively new, and as such more powerful and widespread regulatory controls are now being introduced, although the US still does not have consistent requirements for how data should be handled. Fundamentally, meeting a population’s needs today, and planning for them tomorrow, requires the data of individual people to be analysed. Personal data must be shared quickly, effectively and all the while protecting individual rights. Data system design must therefore:

• Include all players
• Consider cultural constraints
• Keep out bias
• Ensure the right words and phrases are used
• Focus on the ‘so what’, why does it matter?

“Every single thing we experience can be captured as data”

Even the most mundane moments in our daily lives leave a digital footprint, we shed data everywhere. But when does ‘my’ data become public, or the property of the software developer or the service provider? VR headsets collect ephemeral data that is analysed and applied for that one end user, but if that data is assumed to fall under GDPR the potential to use it for positive outcomes is severely limited. For example, should authorities be notified if content viewed and generated is illegal or harmful? And what if that chip can detect if the user is having a stroke, is that data classified as ‘health’ data? Can it be used to alert the individual to their medical emergency without contravening legislation? What if your mouse clicks can detect the early stages of Parkinson’s? Should you, could you, be told?

“If you’re treating this data as health data, then they have a very different set of regulatory constraints. HIPAA isn’t going to regulate those because it’s not a health care provider or a health insurer”

Piercing the veil

The conflict between personal protection and public good is everywhere, and Dr Graeden believes that some new data laws will create problems for federated learning. Legislation has clear boundaries (speed limits, blood alcohol levels) whereas science deals in spectrums, probabilities and unknowns.

Deleting an individual’s personal data from the model breaks the system, contradicting what regulators are trying to achieve. The solution is to prioritize outcomes, not processes – it doesn’t matter whether you write the rules with a pen and paper, or with AI, as long as you write the rules. Expanding the framework by setting gradients of data availability affords protection for individuals, whilst making data available that informs better decision making for public bodies.

“Data is nothing more, nothing less, than an abstract description of our world. A useful and powerful language that can tell us things that other languages don’t”

Data can no longer exist in siloes if it’s to be useful to society

There is now a healthy global appetite for the discussion around data, thanks in the main to two recent developments:

• Covid gave us huge amounts of data about mortality levels, vaccination rates, hospitalisation trends – all of which were in the public consciousness every day
• AI and ChatGPT – articles and debates about the pros and cons are everywhere, discussion is not just in the scientific community

The key challenges now for data scientists are expectation management and communication – we need to be clear about aims and specific about context, as well as knowing what to leave out to avoid overwhelm and misunderstanding. Unfortunately, scientists are not always great communicators (using complex terminology and detail, rather than common parlance and generalization) as Covid demonstrated:

• Did having a vaccine mean you wouldn’t get sick? Or just less sick?
• ‘Everyone should wear a mask’ became ‘wear a mask if you can’. This was due to limited supply, but it appeared that the science was not clear

“The scientific approach means you never have an answer… we are trained as scientists to focus on the fact that we don’t know”

In fact, the only answer is that the right data, used consistently and communicated clearly, will always allow us to be prepared, not reactive. To make decisions for the public good that protect every individual.

You can find out more about the common language of privacy in our Rosetta Stone eBook.

You can also watch Tessa’s fascinating podcast with Dr Graeden below.

The post Making complex data available for the benefit of society appeared first on Calligo.

In Calligo’s latest Beyond Data podcast, co-hosts Sophie Chase Borthwick and Tessa Jones are joined by Alexander Visheratin, Artificial Intelligence Engineer at Beehive AI. Here we explore some of the episode’s highlights; the importance of Natural Learning Processing (NLP) and the pros and cons of output produced by examples like OpenAI’s ChatGPT-3.

“It can do anything, because it was trained on everything”

NLP models like ChatGPT are changing the way we search for data online. But if you average everything, the output will necessarily be average. And we have questions:

• How ethical is the learning data that feeds these models, and how ethical was the process of collecting it?
• How can global models be policed and regulated within individual countries?
• What is the potential for small and specific training datasets to be manipulated by humans in a way that will limit and create biases in the algorithms?
• Is it a ‘bug’ when a prompt doesn’t give us what we wanted? What we ask for is rarely what we actually get.

Confidence or competence?

One major drawback of the NLP process is that many models stopped learning at the turn of the decade, which as Alexander highlights, can easily lead to incorrect information being generated. “I asked one of the large models, ‘who is the president of the United States?’ and it answered very confidently, Barack Obama.” That confidence is interesting, because as humans we are predisposed to trust information that is given to us clearly and directly, with no hint of doubt.

Also, NLP models are built to prove or agree with the task given to them, and they sound so plausible. Alexander shares a specific example of Chat-GPT providing convincing output that could easily persuade someone unfamiliar with the facts.

“Andrew Ng, who is an Adjunct Professor at Stamford University, asked Chat-GPT to prove that CPU is better than GPU for deep learning. It was very confident and created a long paragraph of text proving it. Then he asked it to prove that some more primitive way of calculating is better than CPU, and it again provided very confident paragraph of text. He ended up basically ‘proving’ that an abacus is better than GPU for deep learning.”

In this age of misinformation, there is huge potential for NLP to spread misleading (or downright false) information very quickly to large audiences. ‘Facts’ which then become accepted, magnified and transmitted further.

Taking liberties with artistic license

There are obvious intellectual property issues when it comes to NLP and art generation. Asking an AI tool to create a piece in the style of a named artist will generate convincingly similar work. But if this output contravenes the artist’s morals or political views for example, it is easy to see how discomfort (and possibly even legal challenges) could follow. Conversely, when original artwork is produced that has been generated from hundreds of command iterations to finesse exactly the output required, can it still be seen as ‘art’? Is it the work of the individual using the AI tool, or the tool itself? But is this any different to the great works credited to Michelangelo that we know were produced in part by his students? Is the value of NLP in art actually more as an idea generator, a source of inspiration for the artist rather than the end point?

Alexander believes that creatives shouldn’t be afraid of natural learning. “I think NLP is more of a supplement, a good supplement, because it allows us to be more creative, pushing forward, advancing. It’s not like a replacement at all, it’s more like a co-worker or a supplemental ghost writer almost.”

Guard rails contain or keep out discriminatory language?

OpenAI were very upfront when ChatGPT first launched about the fact that the model would not allow misogynistic or racist material to be produced. Yet the very nature of the learning process saw AI models scraping huge amounts of learning data from the internet, much of which would inherently be of questionable bias and tone. Thus, what these models are drawing from as ‘normal’ is very much not.

“What Chat-GPT doesn’t allow, it feels like it doesn’t allow not because of how it was trained, but because of the huge amounts of guard rails that OpenAI built around it. So, they basically caged this model into all these sorts of limitations about stuff that it shouldn’t allow. But if you can get past these guard rails and into the model itself, it still has all these biases, like race, gender, all this stuff. It still has it, but they just try their very best to limit the way it can show it. Chat-GPT is essentially a celestial bureaucrat!”

NLPs provide assistance, not autonomy

Going forward, combining NLP output with factual SEO-sourced content feels like best practice when using AI tools. Alexander points out that this is quicker than finding the information yourself too and gives us the opportunity to validate what the models generate. Ultimately, he believes that directed and federated learning have fantastic potential, as long as we remain mindful of the risk of reverse engineering and privacy breaches. Using NLP as part of the solution, not the source of the only answer.

If you’d like to discuss the benefits of using Natural Learning Processing in your organization, please contact Tessa Jones to find out more.

You can also watch the fascinating podcast in full below.

The post Unlocking the power of AI and Natural Learning appeared first on Calligo.

In our latest Beyond Data podcast, co-hosts Sophie Chase Borthwick (our Data Ethics & Governance Lead) and Tessa Jones (our Chief Data Scientist) invited Tomer Elias, Director of Product Management at BigID, to discuss how AI bias affects the LGBTQ+ community.

Here we explore some of the episode’s highlights – although you can also watch the full episode here.

Why is there bias?

When building an AI algorithm or AI solution, it is crucial to make sure it’s based on data sets that are both unbiased and diverse and, in terms of the LGBTQ+ community, this often falls short. Whatever the sector – work, health, entertainment – all will be subject to bias if the LGBTQ+ community is not taken into consideration when an AI solution is being created.

For Tessa Jones, one of the barriers to collecting sufficient data is that people might be reluctant to share information about their sexual orientation or their gender journey – particularly if they don’t know how this personal data will be used. Sophie Chase-Borthwick agrees that it quickly becomes a catch-22 situation:

“The biases that make you nervous of disclosing information are the very reason that you need to disclose said personal information in order to prevent bias and improve.

Knock-on effects

Drawing on his experience as a board member of an organization that supports LGBTQ+ employees, Tomer Elias explains how candidates are being let down by recruitment AI solutions and that the consequences are significant.

“A lot of people in the LGBTQ+ community are unemployed and that’s not because they’re lacking the professionalism and passion.”

Meanwhile, medical advances in the LGBTQ+ community are constantly evolving, and many algorithms do not take these changes into account.

“People who are transitioning are not getting the right treatments because the treatment providers are not well educated about it and the data is not diverse enough,” explains Tomer.

Tessa also raises the issue of health apps that require a user to state whether they are male or female.

“Even though the equations could be written differently to how you use different input, they’re just not and that means, you either have to pretend you’re something different or just not use that tool.”

Potential of AI to help overcome bias

While AI bias is clearly affecting the LGBTQ+ community, there are innovative ways it can be used to overcome it, too. Such as in recruitment.

“At the initial interview stage, AI could be used to scramble the voice so you would not know if the candidate was male or female or someone who has transitioned,” says Tomer.

He also poses the possibility for AI to help with the retention of LGBTQ+ employees.

“Technology could help employers know that the employee is happy and feels a part of the organization.”

Time to step it up… 

There are already many AI forces for good – including recommendation systems which can help LGBTQ+ people feel more emotionally supported and The Trevor Project that uses AI to predict which callers are more likely to commit suicide to ensure they get help.

Much more needs to be done. But the fact that people are starting to think about AI bias and the LGBTQ+ community is a step in the right direction.

“Now we’re talking about it and people are realizing the actual real-world implications, hopefully more people will feel comfortable expressing themselves and we can close some of that data gap so there is more information for the models to work off,” according to our Data Ethics & Governance Lead, Sophie Chase-Borthwick.

“It’s also super critical that we have diverse AI developers who are knowledgeable about people and bias,” adds Calligo’s Tessa Jones.

To hear more of our fascinating discussion on AI bias and how it affects the LGBTQ+ community, tune in to our latest Beyond Data podcast episode below.

The post AI bias is frequently failing the LGBTQ+ community appeared first on Calligo.

In our latest Beyond Data podcast ‘Global Food Waste – Can AI Offer a Solution?’, we invited Data Science leader Shawn Ramirez to help us explore the global issue of food waste and discuss how AI has the power to make a difference. Co-hosts Sophie Chase Borthwick (our Data Ethics & Governance Lead) and Tessa Jones (our VP of Data Science, Research & Development) steered Shawn to share her insight and examples of where AI is helping combat this prevalent ‘human’ problem. Here we explore some of the episode’s highlights. 

What a waste… 

To say global food waste is a huge problem seems like an understatement. Nearly one third of all food around the world is currently being wasted. Estimates also suggest that 8-10% of global greenhouse gas emissions are associated with food that’s not consumed. In Shawn’s own words, there are some stark facts that are hard to ignore:  

“In the United States about 40% of food is wasted while, at the same time, 40 million people in the US are suffering from hunger, including 12 million children. If we could reduce and redistribute food waste by 15%, we’d actually feed more than half of those hungry people.  And it’s a similar story in Europe where 153 million tons of food is being wasted.” 

Not just about hunger… 

In addition to hunger, food waste is directly connected to all kinds of additional concerns – such as resource conservation, carbon emissions, and climate change. Clearly hugely passionate about the subject, Shawn explains why we all need to commit to change.   

“With our rising population, the situation is only going to get worse and, if we could reduce or redistribute that food waste, we could have a massive global impact.” 

Where is it happening? 

In the US, 60% of food is wasted before it even reaches the consumer. In Europe, 55% to 60% occurs in consumer households. Whereas in developing countries, most waste happens during agricultural production. Where you are in the world’s supply chains can make a big difference. 

What part can AI play? 

It’s increasingly clear that AI can and is starting to play an important role in turning the tide on food wastage. Exciting hi-tech innovations are enabling more sustainable farming – such as AI-enabled monitors, computer vision, remote sensing, as well as robots. Shawn highlights how this technology is revolutionizing vertical farming. 

“We are now seeing single vertical farms that produce the same amount of fruit and vegetables as an 80-acre farm and they are using 97% less water.”   

A Swedish company is transforming disused office buildings into autonomously controlled greenhouses and a company in Singapore has created the world’s first low carbon hydraulic water-driven vertical farming system.  

Throughout the supply chain, AI is becoming an indispensable planning tool. And Shawn has seen this first-hand, thanks to her time at Shelf Engine – an end-to-end grocery ordering solution using advanced AI. 

“We worked with grocery stores, using inventory simulations to optimize the freshness of food by predicting customer demand…Connecting data across the supply chain facilitates better informed decisions.”   

Knowledge is power

Then there’s a need for efficient and effective monitoring of what’s actually being wasted – something AI now has the capability of doing in granular detail. 

“AI-powered garbage cans equipped with weight sensors, cameras and computer vision have the ability to recognize and track the amount and type of food we’re throwing away.”  

As well as in the home, these can be used in restaurants, hotels, and other businesses – enabling people to think carefully about their waste, while helping companies effectively monitor and understand what’s being thrown away and when.   

You may have heard of Ikea partnering with Winnow Vision AI to track kitchen waste using computer vision technologies. Well, Ikea then used this data to implement changes resulting in a saving of 20 million meals. That amounted to 40,000 tons of carbon dioxide.  

Food for thought

The US Department of Agriculture has set a target of reducing food waste by 10% within the next decade. To achieve this, Shawn believes education in the capability of AI is the next vital step. 

“We want to see more organizations thinking about the food that they waste and realizing how they can make a massive difference by adopting different AI technologies.” 

To hear more of our valuable discussion on how AI has the power to reduce food waste, tune in to our latest Beyond Data podcast episode now. 

The post Global Food Waste – Can AI Offer a Solution? appeared first on Calligo.

Artificial Intelligence’s ability to augment and support progress and development over the past few decades is inarguable. However, when does it become damaging, contradictory even? In our latest Beyond Data podcast AI’s Climate Jekyll & Hyde – friend and foe, Tessa Jones (our VP of Data Science, Research & Development) and Sophie Chase-Borthwick (our Data Ethics & Governance Lead) discuss exactly this with Joe Baguley, Vice President and Chief Technology Officer, EMEA, VMware.

Our speakers explore the multifaceted topic of energy consumption and AI – from whether all applications are equal for energy consumption (or reflecting if there are some ‘better’ than others), to creating visibility and responsibility of energy consumption for all stakeholders. Here we try to give clarity to some of the grey areas that were discussed.

Should we consider all applications equal?

“AI and machine learning are about huge things, huge data sets, huge computation actions … all of those have huge implications in terms of energy,” Joe observes, before dropping in hugely sobering stats such as the total annual energy consumption of bitcoin being the same as Norway. Even when considering the often-touted argument of 57% of the energy for bitcoin mining using renewables, Joe counters: “But those renewables could have been used for something else, right? Those solar panels… and those hydropower stations and those wind turbines, we could be using them for something else.”

This raises the ethical question of whether there should be stricter governance, standards, and precedent set on more ‘moral’ applications for their energy consumption. Should we be more closely considering the difference in energy consumption between server farms that support minimizing food waste versus those that are focused on mining digital currency, for example?

“Is there an opportunity for [greater] regulation?” Tessa ponders. Would this regulation help challenge the current status quo for all applications’ energy consumption being considered equal? While Sophie observes: “We’ve had certain European nations start to put rules around data center expansion, where you’re allowed and not allowed to build because of the capacity there, which isn’t regulating the use of it. But it does have that knock-on effect that if you literally can’t build the data center support, you have to start thinking about other ways to build [models].”

When considering Sophie’s point on alternative ways to build models, Joe notes: “We’re using AI to deal with the symptoms, but maybe there’s some better ways we could be using AI to deal with the cause as well”.

And this all raises the next question – who should ultimately be making these ongoing moral calls for the environment and energy usage?

Embedding Environmental, Social, and Governance (ESG) by design

Environmental, Social, and Governance (ESG) is shorthand for a framework that helps stakeholders understand how an organization is managing risks and opportunities related to environmental, social, and governance criteria. Our speakers untangle the idea of ESG and how companies could use it to help triage the different applications they use.

Joe asks: “Is there an ESG-led marketing opportunity here? Your AI might be the same as my AI, but my AI is better from an ESG perspective. They both get the same results at the same time for the same cost, but this one’s better from an ESG perspective, in terms of sustainability, in terms of social good, in terms of environmental.”

By placing more emphasis on ESG as the criterion for measuring impact and success, it could help with embedding sustainability in the heart of the application’s deployment, rather than a siloed approach. Sophie agrees: “We have privacy by design, we have security by design. Why not have ESG by design?”

Following on from this thought, our speakers consider the cost implications of AI and ESG with Joe observing, “There’s a lot of businesses right now that can’t afford AI because it’s expensive…but I believe they will come to a tipping point where they can’t afford not to”.

Are we over-prioritizing accuracy?

“There’s a hyper-focus on the accuracy,” according to Tessa. “It ends up not even being about the motivation for green, it’s a motivation for fast training, fast tuning. Unfortunately, it’s how most data scientists are motivated; be faster without having to compromise their accuracy.”

Often, the increase in accuracy can be mapped on a logarithmic graph. Good gains at first, but quickly tapering off to minimal increase. Is it useful to be that much more accurate, often by points of a decimal? “Some are good, more must be better … people just keep going, as opposed to saying actually good enough is good enough,” Joe summarizes.

Instead of chasing marginally better accuracy each time, we should be considering the application in a holistic view. The increase in accuracy might be 0.01%, but would cost heavily for energy consumption – is it worth it? Should we be better at exposing these costs more vigorously throughout a team so everyone can feel more empowered and have the visibility to interrogate more closely?

To hear about how our speakers untangle these controversial questions and more, tune in now to Beyond Data podcast episode 3: AI’s Climate Jekyll & Hyde – friend and foe.

The post The dark side of AI energy consumption – and what to do about it appeared first on Calligo.