Truth rests between the two extreme views.
Artificial intelligence (AI) encompasses a technology spectrum that aims to replicate our thinking processes or, at least, the observable outputs. It may extend to recognising objects and shapes, translating languages or deriving insights from impossibly big datasets, just for starters. And it’s inserting itself into the very weft and warp of business and society at every level. Soon, we may wonder how we ever lived without it.
‘Generative AI’ such as ChatGPT (aka a ‘transformer’) that creates text like program code and HTML or Midjourney that draws artistic images from user queries (‘prompts’) seemingly creates content out of thin air like magic. And while it may chill the collective spine of human creatives — AI art has already won awards — such technologies boost productivity, validate approaches and raise fresh perspectives or avenues of inquiry.
Joshua Morley said we should think of them like a sidekick or ‘copilot’ (as Microsoft calls its ChatGPT-based product).
Approach with caution,” advised Morley, Akkodis National Solutions and Practice Lead for IoT and Digital Twins.
“Slamming the shutters down is entirely the wrong move. We’re facing a new reality where generative AI will perform tasks like taking meeting minutes, drafting long-form emails or even writing code.”
Asking the AI Agent to create the template, code stub or a memo draft is a quick productivity win in the same way downloading a PowerPoint template or working in an application like Canva is, said Morley.
In fact, Akkodis in Australia has already begun mobilising a cross business unit “Responsible AI Council” chaired by Morley. When asked about the council Morley said: “Our CEO rightfully pointed out, ‘no single person has all the knowledge required to implement AI responsibly’, By forming this cross business unit council with specific guiding principles including ‘ethical usage’, ‘privacy and security centred’ and ‘human oversight and control’, among others, we can ensure we are steaming ahead and embracing the efficiencies and opportunities AI has to offer within the bounds of good governance and data privacy.”
Furthermore, Akkodis Australia is already deploying AI Agents to augment business operations and provide efficiencies. The Tech Talent business recognised the value in using Generative AI text tools and in a pilot program has demonstrated the ability to drop the time taken from role request to job posting down to a mere 10 minutes, an eightfold decrease in time taken.
AI looks great but what’s the catch?
Know that generative AIs often get things wrong — and may even lie to you, what is euphemistically called an ‘hallucination’. So always double-check what they say and maintain a critical mindset towards their output (fact-checking with different information sources is critical).
“I would never see it writing something to completion that you would submit without human review — at least not for the immediate future, anyway,” Morley said.
And Morley cautioned that AIs may also steal your data — however unwittingly.
“Identify and understand whichever generative AI tool you’re using. Know what their policy is around data security and privacy, and change those settings so that you’re not inadvertently releasing confidential information out into the world.”
Why generative AI won’t arrest crooks any time soon
Science fiction authors such as Isaac Asimov and Philip K. Dick have long written about synthetic investigators helping police bring criminals to justice.
Jeremy Dennis, who collaborates with law enforcement agencies on their information systems and participates in Akkodis’ AI ethics workgroup, said generative AI doesn’t meet minimum legal thresholds.
“I can think of nothing worse from a humanities perspective,” said Dennis, Akkodis Global Analytics Lead.
“As soon as we rely on an AI to generate content in the context of law enforcement, we end up in a situation where — and quite rightly — we cannot accept that content in a court of law.”
“If somebody is arrested and charged with a criminal offence because an AI predicted or generated content, is an extremely dangerous scenario”
Dennis said the “stakes are so high; we’re talking about people’s livelihoods”.
“You need to be really sure that proper process has been followed, and I don’t see a place for generative AI in this context.”
Which isn’t to say that crooks get a free pass. Big data and analytics are pivotal to bringing criminals to justice.
Named after the antagonist in The Usual Suspects, ‘Söze’ is an Akkodis platform to help investigators analyse vast amounts of data on a multitude of devices belonging to people under investigation.
“Modern investigations generate vast amounts of digital evidence. Enabling investigators to rapidly analyse this data and discover avenues of potential interest is vital to solving crimes more quickly, and ultimately to reducing harm to the community.”
Söze cuts turnaround from weeks to seconds by ingesting any of hundreds of known file formats from almost any device, enabling investigators to start analysing and forming opinions. This could include tracking where people were on a map, who they were with, whether they communicated with each other and their relationships. And it may reveal previously unknown witnesses, he said.
Big data and analytics brings justice to cold cases
AI is also reopening cold cases for closer scrutiny, raising hopes that people could be brought to justice sometimes decades later.
Not only does Söze slash cold case investigators’ workload when information is metatagged and indexed, it also answers multi-faceted questions easily. Such ‘geotemporal’ questions as: “Who was in this area at the time” or “Was this person ever near this other person” are now possible once Söze has raw data such as sergeants’ reports and witness statements, Dennis said.
“We get powerful insights and pick up things that were missed because the data was scattered. That opens investigative avenues and that’s really important.”
Who are you? A conversation with ChatGPT
> Explain as if you were speaking to a non-technical business leader (eg CEO, CFO) what are the different types of artificial intelligence, how they work, and where they’re typically to be found?
>> AI encompasses a range of technologies and approaches that aim to replicate or augment human intelligence.
- Expert Systems/Rules-Based Systems
- Machine Learning
- Neural Networks
- Deep Learning
- Reinforcement Learning
- Natural Language Processing (NLP)
- Computer Vision
- Generative Adversarial Networks (GANs)
- Fuzzy Logic
- Swarm Intelligence
- Evolutionary Algorithms
- Knowledge Representation and Reasoning (KRR)
- Bayesian Networks
- Expert Systems
Narrow AI is prevalent and focused on specific tasks, while general AI remains a goal for future development. Machine learning and deep learning are key techniques within AI that enable systems to learn and make predictions based on data.
These technologies are often found in industries such as healthcare, finance, retail, and manufacturing, where they can enhance productivity, efficiency, and decision-making. However, implementing AI can also require significant investments in infrastructure, data quality, and technical expertise.
Visit Akkodis Australia to learn how the global digital engineering powerhouse can speed your energy and clean technology transformation.