AI: Friend or Foe?
In the tech world few topics are hotter than Artificial Intelligence (AI). Is it our friend? Or is it our foe?
The debate over its merits, limits and mooted dangers has seen intellectual and technological heavyweights clash, amongst them X’s (formerly Twitter) supremo Elon Musk and Meta’s Chief AI Scientist, Yann LeCun.
Last year Musk told the conservative political commentator Tucker Carlson that “AI is more dangerous than, say, mismanaged aircraft design or production maintenance or bad car production, in the sense that it is, it has the potential — however small one may regard that probability, but it is non-trivial — it has the potential of civilisation destruction.”
LeCun, whose career in machine learning and AI spans the best part of four decades, has little time for such dystopian theories, arguing that “people are exploiting the fear about the technology” and proponents of such ideas from within the tech sphere are “naïve” or “attention seeking.”
In an interview with Wired’s Stephen Levy he stated that: “They don’t realize that AI actually mitigates dangers like hate speech, misinformation, propagandist attempts to corrupt the electoral system.” In 2022 he claimed AI pre-emptively removed 95% of hate speech posted on Facebook - before it could be seen.
Talking to venture capitalist Henry Stebbing – as reported by Business Today - he brusquely dismissed Musk’s prophecy of doom, reasoning that it’s based on a false premise, this being that when you “turn on a super-intelligent AI system, it’s going to refine itself to be even more intelligent than humans and the world will be destroyed.
“That’s completely ridiculous because there is no process in the real world that is exponential for very long. Those systems will have to recruit all the resources in the world. They would have to be given limitless power, agency.”
AI-powered innovation and growth
AI is here and it’s here to stay. It delivers untold benefits to our daily lives, powering game-changing innovations such as Google Maps’ AI-driven Immersive View for Routes facility.
By creating a digital model of the world this offers multi-dimensional views of planned routes by car, bike and on foot and, by drawing on historical data, can even provide insights into likely weather and traffic conditions on a given day and time.
You’d also be hard-pressed to find any analyst who thinks that AI will not promote economic growth.
Arguing that technological disruptions are fostering what he calls “flash growth”, David Shiel, Professor of Practice, AI and Innovation, Imperial College Business School, has stated that AI could fuel a 10% increase in global gross domestic product by 2032. In real terms that’s £10 trillion.
International professional services giant PWC is even bolder. Its Sizing the Prize report points to a €15.7 trillion - circa £12.6 trillion - global economic boost by 2030.
In achieving this the challenge will be to promote responsible application of AI through a mix of governmental and international regulation and the inclusion of guardrails by developers.
The defining technology of our time
Microsoft, which has published its own blueprint for the safe and transparent development and deployment of AI in the UK, emphasis the significant benefits the technology is bringing –and will bring – to private and public sector organisations.
The tech giant, which regards AI as “the defining technology of our time”, cites how the NHS is improving patient care and experiences, reporting how, by replacing labour-intensive manual analysis of organ scans, AI has empowered oncologists at Addenbrookes Hospital to cut waiting times for cancer patients. This has accelerated treatment planning by around two-and-a-half times, meaning patients get seen more quickly.
The company also cites the productivity and economic gains to be made, pointing to International Data Corporation research that for every dollar invested in AI, companies secure a $3.50 return. That return was also realisable in just 14 months.
The same research quotes Russell Johnson, Marks & Spencer’s Chief Data Scientist, who reports huge real-world benefits at the iconic retailer, stating that: AI has been a game-changer for Marks & Spencer, delivering exceptional value across our enterprise and personalisation initiatives.
“We have seen a tenfold return on our enterprise investments, generating seven figures in savings and efficiencies. On the personalisation side, we have unlocked eight figures in incremental revenue thanks to AI-powered recommendations and offers, which has exceeded our expectations on the impact AI can have on our business.”
Closer to home Avoira clients are reporting how our AI-driven Xdroid customer experience and speech analytics solution is powering huge productivity gains whilst enhancing customer service.
For example, contact centres are benefiting massively from Xdroid’s ability to capture, transcribe and analyse every call. This removes the need for manual oversight by automating and evidencing adherence to scripts and regulatory requirements. This saves a tremendous amount of time whilst yielding comprehensive rather than sampled data to inform training needs and enhance service.
We should not forget too that, whatever we do, AI is already heavily embedded in our everyday lives, offering a helping hand with everything from checking our grammar to determining, in real time, the fastest route to our chosen destination.
It’s here. It not going anywhere. Quite the opposite, it’s going everywhere. The key issue now then is how future AI innovations are mapped and deployed for the good of all. This will be determined both by state and territorial regulators – the EU this March enacted it its own Artificial Intelligence Act – and developers themselves.
Objective Driven AI
Meta’s LeCun is an advocate for what he calls “Objective-Driven AI” (ODAI) and the benefits it will bring within and without Meta’s sphere.
An ODAI model does what it says on the tin – fulfils specific goals defined by its human developers. This is, or will be, achieved by learning not just through the text-gobbling Large Learning Models (LLMs) upon which generative AI depends, but video data and sensory input from the real world.
Development of ODAI models will be transformative but, crucially, through the inclusion of “guardrails”, mean that AI remains our friend. These will and already do ensure a model remains within its remit, delivering only pre-determined outcomes and preventing unintended consequences.
The guardrail concept is one embraced by Xdroid which has delivered such hugely impressive quantitative and qualitative benefits to Avoira clients.
The company balances the immense potential AI offers in revolutionising operations, products and services, with the need to ensure that it is developed, deployed and utilised ethically and responsibly.
Ethical AI in action
In developing AI models it’s important to safeguard against things like unconscious biases, inappropriate collection or use of data and the misuse for malicious purposes of LLMs which inform features such as Xdroid’s sentiment detection.
Guardrails are very much in evidence with the platform trained specifically to perform a defined task and eliminate AI-related risks.
So, for example, LLMs necessarily draw from huge pools of real-world data which can incorporate embedded biases. To mitigate against this, Xdroid firstly ensures that the data used to train a model represent appropriate diversity and scrutinises it before it is processed. De-biasing techniques are additionally employed during model training and for fine-tuning.
The process continues through continuous monitoring, using evaluation metrics and human analyses. De-biasing techniques are additionally employed during model training and for fine-tuning.
In ensuring fairness and equity, bias detection, and mitigation measures safeguard against discrimination against race, gender, ethnicity, age, socio-economic status and other protected attributes.
Xdroid is also able to recognise contextual nuances, a power which distinguishes it from more rudimentary AI platforms. Contextual understanding is not only vital to the future development of Xdroid models but the wider application of AI. Lack of it is a key criticism of generative AI tools which can, as a result, provide outputs which misinform or mislead.
Principle into practice
At Xdroid a key development principle is to prioritise the well-being and interests of individuals and communities, with AI systems designed to one, augment human capabilities; two, enhance user experiences; and three, make a positive contribution to society.
So, how might those three design features translate into practice? The bespoke application by Avoira of Xdroid at Principal Insurance gives some insight.
As remains common practice for many insurance brokers, Principal previously relied on manual analysis of randomly sampled agent calls. “Not only is that a time-consuming process, but an arbitrary one,” Principal’s Head of Compliance & HR, Sharron Titterington told us.
Now, Xdroid augments human capability by capturing, transcribing and analysing every single call. In addition, with the model trained to eliminate bias, it provides impartial reporting of agent performance and caller sentiment.
User experience has proven to be significantly enhanced, whether the user is a senior manager or call agent. “Through its agent and management dashboards, Xdroid offers continual assessment of compliance and wider performance. Agents like that feedback and it incentivises them to improve, to beat their score,” reports Sharron.
And societal good? A key component of Principal’s Xdroid solution is a Vulnerable Customers’ Identification Tool. This was developed in partnership with our CX team and saw the model being trained to flag callers who may be subject to any of a wide range of vulnerabilities – from low income through domestic violence, depression and suicidal tendencies.
This allows Principal to investigate and, where appropriate, offer guidance and support to callers. It also evidences the broker’s compliance with the Financial Conduct Authority’s stringent Consumer Duty regulation which, rightly, places great emphasis on securing fair treatment of vulnerable customers.
In addition, deployment of a Customer Sentiment Tracker enables oversight not just of customer satisfaction levels throughout a call, but can identify hostile or abusive content directed at agents, making it an excellent tool for supporting staff wellbeing.
We were able to deliver such benefits because of our and Xdroid’s commitment to transparency and a human-centric approach.
User-centred design means we always prioritise the needs, preferences and experiences of our clients, working in partnership with them through a thorough discovery, testing and feedback process to ensure the AI systems meet their needs and enhance experience.
At a higher level, every Xdroid model enables maintenance of human oversight and the ability to intervene in an AI-driven decisions.
Transparency is important not least because the clients we serve cannot be expected to possess the knowledge and understanding of AI models that our CX experts have.
Accordingly, as well as making available the design, architecture, and parameters of our AI models, we don’t hide behind jargon. Instead, we ensure that all stakeholders can comprehend the reasoning behind AI-driven decisions and both the limitations and capabilities of the technology.
Making a friend
The heat inherent in the debate over the merits or otherwise of AI is going to remain intense for some time yet.
The big tech companies are in a race for the prize and as such offering competing products and conflicting messages.
At the same time regulatory authorities are grappling to determine the most appropriate frameworks to guide and enforce responsible development.
It is that last – responsible development – that is going to be key. As with any technology, it’s how it’s used those matters.
I’m proud of Xdroid’s pro-active, best-practice stance on the ethical development and application of AI. It enables Avoira to deliver solutions that are as disciplined and effective in meeting our clients’ specific goals as they are fair and safe.
To find out more about Avoira, get in touch with their team!
